Protective Mechanisms of Respiratory Tract Streptococci against Streptococcus pyogenes Biofilm Formation and Epithelial Cell Infection

Abstract

Streptococcus pyogenes (group A streptococci [GAS]) encounter many streptococcal species of the physiological microbial biome when entering the upper respiratory tract of humans, leading to the question how GAS interact with these bacteria in order to establish themselves at this anatomic site and initiate infection. Here we show that S. oralis and S. salivarius in direct contact assays inhibit growth of GAS in a strain-specific manner and that S. salivarius, most likely via bacteriocin secretion, also exerts this effect in transwell experiments. Utilizing scanning electron microscopy documentation, we identified the tested strains as potent biofilm producers except for GAS M49. In mixed-species biofilms, S. salivarius dominated the GAS strains, while S. oralis acted as initial colonizer, building the bottom layer in mixed biofilms and thereby allowing even GAS M49 to form substantial biofilms on top. With the exception of S. oralis, artificial saliva reduced single-species biofilms and allowed GAS to dominate in mixed biofilms, although the overall two-layer structure was unchanged. When covered by S. oralis and S. salivarius biofilms, epithelial cells were protected from GAS adherence, internalization, and cytotoxic effects. Apparently, these species can have probiotic effects. The use of Affymetrix array technology to assess HEp-2 cell transcription levels revealed modest changes after exposure to S. oralis and S. salivarius biofilms which could explain some of the protective effects against GAS attack. In summary, our study revealed a protection effect of respiratory tract bacteria against an important airway pathogen and allowed a first in vitro insight into local environmental processes after GAS enter the respiratory tract.

INTRODUCTION

Streptococcus pyogenes (group A streptococci [GAS]) belongs to the most important bacterial species causing purulent respiratory tract and skin infections in humans in more than 720 million patients per year (1, 2). Many of the GAS virulence factors involved in these infectious processes, as well as the underlying circuits controlling their expression, have been identified in the recent years (3–9).

The administration of β-lactams is the treatment of choice to fight purulent GAS infections. Despite this appropriate therapy, 5 to 10% of patients suffer from recurrent episodes (10). This phenomenon could be associated with the ability of GAS to adhere to and, more importantly, internalize into eukaryotic cells (3, 11). Unfortunately, a high capacity for internalization is apparently associated with resistance to macrolides (12), preventing this intracellularly active antibiotic from being a true therapeutic alternative, especially for recurrent GAS infections. Therefore, new options would be welcome to combat such infections, and, in final consequence, the spread of the bacteria from such carrier patients in a nonepidemic setting.

Probiotics, defined as “live microorganisms [that] when administered in adequate amounts confer a health benefit on the host” (FAO/WHO, 2002) could serve as a therapeutic alternative (13). Beneficial roles of probiotic bacteria have already been reported for treatments in the oral cavity (14). Thus far, only a few products became commercially available, e.g., the Streptococcus salivarius strain K-12, which is sold as BLIS lozenges in New Zealand (15), and freeze-dried powders of Lactobacillus rhamnosus plus Bifidobacterium casei and Enterococcus faecalis, which are sold as Symbiolact compositum and Symbioflor, respectively (16, 17). Like other probiotics these have three major activity routes. First, using appropriate adhesins, detergents, bacteriocins, or quorum-sensing inhibitors, the probiotic microorganism can outgrow and permanently replace pathogens. Second, the host defense mechanisms are stimulated or modulated in favorable ways to eradicate persisting pathogens and better resist future pathogen exposure. Third, after the application of antibiotics or antiseptics, void spaces on human surfaces could be filled by robust harmless bacteria until the resident microbial biome is reconstituted.

GAS are transmitted between humans by direct contact or by contaminated airborne droplets. Therefore, the skin and the upper respiratory tract are the primary GAS contact sites on new human hosts (18). Both anatomic sites are physiologically inhabited by several dozen (skin) to several hundred (respiratory tract) different bacterial species (19–23), of which the majority can only be demonstrated by molecular techniques (24, 25). Many of these species have probiotic potential that could be exerted by biofilm formation on anatomical sites serving as targets for pathogenic bacteria. Several recent studies have shown that dairy bacteria such as Lactobacillus helveticus, L. rhamnosus, L. oris, and L. reuteri, as well as other oral bacterial species, can protect the pharyngeal mucosa from GAS attack (26, 27). Thus far, it is unclear whether this defense mechanism in vivo involves biofilm growth.

In turn, the pathogens could build their own biofilm or integrate into existing ones. GAS form biofilms in vivo and in vitro (28–30). Several GAS virulence factors have been suggested to be involved in biofilm formation, including M protein, antigen I/II family polypeptide AspA, SpeB production, and the GAS pilus (31–36).

Several issues associated with the protective biofilm hypothesis have not been resolved. (i) Can probiotic bacteria interfere with the biofilm formation capacity of GAS? (ii) Are GAS able to invade and establish within existing probiotic biofilms. (iii) Can probiotic bacteria protect epithelial cells from infection with GAS. (iv) Finally, what are the underlying mechanisms of these interactions with regard to transcription and function levels. Consequently, in the present study we used the well-studied BLIS S. salivarius K-12 strain and two Streptococcus oralis strains (DSMZ reference strain DSM20627 and patient isolate 4087) as representatives of the predominant upper respiratory tract species to test their potential protective capacity against GAS biofilm formation and host cell infection.

MATERIALS AND METHODS

Strains and culture conditions.

S. pyogenes M49 strain 591 (37) and M6 strain 616 (clinical isolate, Centre of Epidemiology and Microbiology, National Institute of Public Health, Prague, Czech Republic), S. salivarius strain K-12 (BLIS Technologies, Ltd., Dunedin, New Zealand), and S. oralis strains DSM 20627 (German Collection of Microorganisms and Cell Cultures [DSMZ]) and 4087 (isolate from a healthy person [the present study]) were cultured in brain heart infusion (BHI) or on Columbia blood agar plates containing 5% sheep erythrocytes at 37°C under a 5% CO₂–20% O₂ atmosphere. Eukaryotic laryngeal epithelial cell line HEp-2 (American Type Culture Collection) was cultured in Dulbecco modified Eagle medium (DMEM) supplemented with 10% fetal calf serum (FCS) in cell culture flasks or 24-well microtiter plates under a 5% CO₂–20% O₂ atmosphere.

Coaggregation experiments.

Bacteria were harvested from overnight cultures by centrifugation and washed twice with coaggregation buffer (tris[hydroxymethyl]aminomethane, 1 mM [pH 8.0]; CaCl₂, 0.1 mM; MgCl₂, 0.1 mM; NaN₃, 0.02%; NaCl, 0.15 M) (38). The suspension of bacterial cells in coaggregation buffer was adjusted to an optical density at 600 nm of 2.0. To assess two-species coaggregation, the respective strains were mixed 1:1 to a final volume of 4 ml and vortexed for 10 s. After 2 h of incubation at room temperature coaggregation was determined by visual inspection.

Biofilm culture conditions.

For biofilm assays, conditions optimized for GAS were used (39). In detail, bacterial overnight cultures in BHI broth were suspended in fresh BHI broth supplemented with 0.5% glucose (BHI-G), adjusted to 10⁴ CFU/ml, and inoculated into 96- or 24-well microtiter plates for quantification of biofilm masses or microscopic analyses, respectively. Alternatively, BHI broth mixed 1:3 with artificial saliva (0.1% Lab Lemco Powder, 0.2% yeast extract, 0.5% peptone, 0.25% mucin from porcine stomach [type III; Sigma], 6 mM NaCl, 2.7 mM KCl, 3.5 mM KH₂PO₄, 1.5 mM K₂HPO₄, 0.05% urea [pH 6.7]) was used as cultivation medium. For one-species biofilms 100 μl of bacterial suspension (10⁴ CFU/ml), and 100 μl of BHI-G were added per well of a 96-well plate. For two-species biofilm cultures, 100 μl of bacterial suspension (10⁴ CFU/ml) of each species were simultaneously inoculated per well. For microscopic observations, biofilms were generated in 24-well polystyrene plates (Cellstar Greiner; Bio-One) with inserted round plastic coverslips (Nunc) or in glass-bottom 2/4-chambers (Nunc). Bacteria were inoculated in ratios as described above to a final volume of 1 ml. Biofilm analysis was performed after incubation for 3 days as standing cultures at 37°C under 5% CO₂–20% O₂ atmosphere. All assays were performed in triplicate (technical replicates) on at least three independent occasions (biological replicates).

Scanning electron microscopy (SEM).

Biofilm bacteria were cultured as described above. After the removal of remaining planktonic bacteria, biofilms on coverslips were washed with phosphate-buffered saline (PBS) and fixed with 2.5% glutaraldehyde overnight. Coverslips were washed two to three times with 0.1 M sodium phosphate buffer (pH 7.3), dehydrated with a series of increasing ethanol concentrations (5 min in 30%, 5 min in 50%, 10 min in 70%, 10 min in 90%, and twice for 10 min in ethanol absolute), and dried with CO₂ by critical point method with a Emitech dryer as outlined by the manufacturer. Dried coverslips were covered with gold to a 10-nm layer and scanned with a Zeiss DSM 960A electron microscope (29).

Confocal laser scanning microscopy (CLSM).

The visualization of S. pyogenes in single- and mixed-species biofilms was done by immunofluorescence staining using a specific anti-S. pyogenes polyclonal antibody together with an Alexa 488-coupled secondary antibody. Hexidine iodine was used for staining all Gram-positive bacterial species. After the removal of planktonic bacteria, the biofilms were washed once with PBS, fixed with 3% paraformaldehyde for 15 min at 4°C, and subsequently blocked with 1% FCS at room temperature for 30 min. The biofilm was washed three times with PBS and incubated for 1 h at room temperature with a rabbit IgG anti-S. pyogenes antibody (1:5,000 in PBS). After being washed with PBS, the second antibody (goat anti-rabbit-IgG-Alexa Fluor 488) was added at a 1:500 dilution in PBS for 45 min at room temperature in the dark. Again, three washing steps with PBS were performed. Hexidine iodine (1 μl in 1 ml of PBS [Invitrogen]) was added for 10 min at room temperature in the dark. The biofilm was inspected with a Zeiss inverted microscope attached to a Leica TCS SP2 AOBS laser scanning confocal imaging system with an Argon laser at 480- and 488-nm excitation wavelengths.

Bacteriocin assay.

The bacteriocin assay was performed based on a modified deferred antagonism cross streak technique on blood agar (40). Potential bacteriocin producers were streaked on blood agar, incubated for 18 h, and subsequently killed with a vapor of chloroform for 4 min. Chloroform vapors were removed by passive ventilation of the agar plate for 10 to 15 min. S. pyogenes strains were streaked across the primary streak and incubated overnight. Growth inhibition and beta-hemolysis inactivation were observed around the intersection of bacterial streaks.

Adherence and internalization assays.

Adherence to and internalization into HEp-2 cells was quantified using an antibiotic protection assay (41). Three different seeding strategies were used for this assay: (i) simultaneous seeding (S. pyogenes and respiratory tract streptococci were added to the HEp-2 cells at the same time), (ii) S. pyogenes first seeding (S. pyogenes was incubated with HEp-2 cells for 1 h prior to addition of the S. oralis or S. salivarius), and (iii) S. pyogenes last seeding (respiratory tract streptococci were incubated with HEp-2 cells for 2 h prior to inoculation of S. pyogenes). Moreover, several modifications of the latter method were performed. (i) DMEM medium and planktonic respiratory tract bacteria were removed prior to S. pyogenes infection, and Hep-2 cells were washed three times with sterile prewarmed PBS preceding the S. pyogenes inoculation. (ii) A transwell system was used, preventing direct contact between the respiratory tract bacteria (upper compartment) and HEp-2 cells (lower compartment). In addition, adherent and internalized S. pyogenes chains were visualized using double immune fluorescence as described previously (42). All assays were performed with three technical and at least three biological replicates.

Eukaryotic cell viability assay.

The live/dead viability/cytotoxicity kit for animal cells (Molecular Probes; Mobitec) was used to determine viability of HEp-2 cells in the presence of S. pyogenes and respiratory tract species. Exposure of HEp-2 cells to the respective bacteria was as described for adherence and internalization experiments, except that HEp-2 cells were cultivated in 24-well plates on coverslips. Visualization and inspection of live or dead HEp-2 cells was done under a fluorescence microscope at 400-fold magnification. In each assay, three randomly chosen microscopic fields were documented as a picture. In order to express the results as quantitative data, living cells (green fluorescence) were counted and expressed as the percentage of all cells (live and dead) visible in each picture.

HEp-2 cell transcriptome analysis.

To compare host cell gene expression profile changes caused by S. salivarius K-12 and S. oralis DSMZ, high-density oligonucleotide microarrays were applied. Two biological replicates of total RNA samples from infected HEp-2 cells were hybridized to Human Genome U133 plus 2.0 array (Affymetrix, St. Clara, CA), interrogating 47,000 transcripts with more than 54,000 probe-sets. Array hybridization was performed according to the supplier's instructions using GeneChip expression 3′ amplification one-cycle target labeling and control reagents (Affymetrix). Subsequently, washing and staining protocols were performed with the Affymetrix Fluidics Station 450. For a signal enhancement, an antibody amplification was carried out with a biotinylated anti-streptavidin antibody (Vector Laboratories, United Kingdom), which was cross-linked by a goat IgG (Sigma, Germany), followed by a second staining with streptavidin-phycoerythrin conjugate (Molecular Probes/Invitrogen). The scanning of the microarray was done with a GeneChip Scanner 3000 (Affymetrix) at a 1.56-μm resolution.

The data analysis was performed using MAS 5.0 (Microarray Suite statistical algorithm; Affymetrix) and probe-level analysis using GeneChip Operating Software (GCOS 1.4), and the final data extraction was performed with the DataMiningTool 3.1 (Affymetrix). Differentially upregulated and downregulated genes from two independent experiments were clustered manually and analyzed for their molecular function using tools in NetAffx analysis center (http://www.affymetrix.com/analysis/index.affx).

Statistical analysis.

Where applicable, statistical analyses were performed using the Student t test. Significant differences were defined when the P values were <0.05.

RESULTS

Cocultures of S. pyogenes and respiratory tract streptococci in liquid culture media.

Only scarce information exists regarding how S. salivarius and S. oralis could influence cultures of GAS. Thus, coculture experiments were performed by growing GAS serotypes M49 and M6 in direct (tubes) or indirect (transwell system) contact with the two respiratory tract streptococci in several combinations. In initial coaggregation experiments, coincubation of the different species was shown to increase the bacterial dispersion in combinations of GAS with S. salivarius and S. oralis DSMZ (see Fig. S1 in the supplemental material), indicating a lower degree of coaggregation in planktonic mixtures of these species. Serial combinations of high to low bacterial CFU counts were used to test whether bacterial numbers play a role in this interaction. From the outcome of these experiments, only meaningful results are shown. As shown in Fig. 1, the direct contact experiments revealed that a high initial CFU count of S. salivarius and S. oralis mixed with a low GAS M49 and M6 CFU count caused the repression of GAS growth. In contrast, low numbers of S. salivarius CFU inhibited the growth of low initial GAS M49 and M6 CFU (Fig. 1A and D).

Fig 1.

Cocultivation experiments with S. pyogenes and respiratory tract streptococci. The figure shows CFU of S. pyogenes M49 (A to C) and M6 (D to F) after 24 h of either coincubation in direct contact (light gray bars) or in transwell assays (dark gray bars) with S. salivarius (Ss) and S. oralis (So) DSMZ strain or strain 4087, respectively. Numbers below the bars indicate the initially inoculated number of bacteria (in CFU/ml). Error bars resulted from three independent experiments.

In experiments using the transwell system, secreted substances from most bacteria did not decrease GAS CFU, except for S. salivarius (Fig. 1). The initial mixture of 10⁶ CFU of S. salivarius/ml plus 10³ CFU of GAS/ml led to a 7-log decrease in GAS CFU compared to untreated controls. Furthermore, the S. oralis strains reduced GAS M49 CFU by 1 log (Fig. 1B and C). No other strain combination or CFU variation revealed significant effects. The effect of S. salivarius on GAS M49 was more prominent than the effect on GAS M6. Apparently, S. salivarius secreted a diffusible substance that exerted a growth inhibition effect on GAS, and the GAS M6 strain is more resistant against its action.

The fitness of the tested respiratory tract species in the presence of GAS strains was also monitored in the direct contact system (see Fig. S2 in the supplemental material). Even mixtures of initially high GAS CFU with low S. salivarius and S. oralis CFU did not prevent the growth of the respiratory tract streptococci in the mixed cultures.

Bacteriocin production.

One possible explanation for some of the results of the coculture experiments could be the secretion of diffusible substances such as bacteriocins. Thus, we tested bacteriocin production by the respiratory tract species using a solid blood agar test system. The potential producers were grown in a first streak and inactivated by chloroform vapor. Hence, any observed effect was caused by chloroform-stable, diffusible substances that were produced by the first streaked strain. The results shown in Fig. 2 revealed that exclusively S. salivarius could kill both GAS strains by bacteriocin production.

Fig 2.

Bacteriocin production. S. salivarius K-12 (A), S. oralis DSMZ (B), and S. oralis strain 4087 (C) were grown on Columbia agar plates overnight and killed by chloroform exposure. Subsequently, S. pyogenes M6 and M49 were perpendicularly streaked across the primary streak and incubated for another 24 h. Growth inhibition zones are visible close to the S. salivarius streak. Hemolysin production is obviously not selectively suppressed in growing S. pyogenes bacteria.

Combined species biofilm phenotypes.

GAS are able to form biofilms in vitro. However, the biofilm phenotype is serotype dependent (29, 39). Here, biofilm formation of GAS M6 (strong biofilm producer) and M49 (poor to moderate biofilm producer) was investigated in the presence of S. salivarius and S. oralis strains to observe whether GAS can integrate into respiratory tract biofilms or GAS biofilms can be disturbed by respiratory tract bacteria.

Initial tests revealed that BHI supplemented with 0.5% (wt/vol) glucose best supported growth of all species (data not shown). Prior to mixed-species experiments, the biofilm-forming ability of all strains was assessed using monospecies cultures. From the safranin assay results and SEM pictures (Fig. 3), it is apparent that all species developed significant biofilm masses except for the GAS M49 strain in accordance with published results (29). Bacterial coaggregation was previously shown to play an important role in biofilm development and bacterial interaction with host cells (43). Therefore, coaggregation capacity of BHI-grown single- and mixed-species was examined. Only the GAS strains revealed an aggregation effect in single-species experiments. In addition, S. oralis 4087 coaggregates with both of the GAS strains tested (see Fig. S1A-C in the supplemental material).

Fig 3.

Single-species biofilms. The biofilm-forming capacity of M6 and M49 S. pyogenes strains, as well as of S. salivarius K-12, S. oralis DSMZ 20627, and S. oralis 4087 strains, was assessed after 72 h of incubation by safranin staining and absorption measurements at 492 nm. Error bars resulted from three independent experiments. The panel below the graph shows SEM images of the biofilms of the respective strains after 72 h of incubation.

After we established the optimal conditions for monospecies biofilms, we investigated the isolates' behavior in mixed-species settings. As documented by SEM and CLSM, biofilms of both GAS serotype strains plus S. salivarius were dominated by S. salivarius bacteria, since GAS were hardly visible in this combination (Fig. 4; see also Fig. S3 in the supplemental material). This finding was supported by the biofilm masses as quantified by safranin staining. The amounts of the S. salivarius/GAS two-species biofilms were almost identical to those of S. salivarius monospecies biofilms but were significantly higher than GAS M6 and M49 monospecies biofilm masses (Fig. 4). Obviously, the observed effects were independent of the biofilm-forming capacity of the individual GAS strain.

Fig 4.

Biofilm formation of cocultures of S. pyogenes M49 (A) or M6 (B) and respiratory tract streptococci. After 72 h of incubation, the biofilms were quantified by safranin staining and absorption measurement at 492 nm. White bars represent the S. pyogenes monospecies biofilms, light gray bars represent the monospecies biofilms of the respiratory tract streptococci, and dark gray bars represent the mixed-species biofilms. Error bars resulted from three independent experiments, asterisks indicate significant differences (Student t test: **, P < 0.01; ***, P < 0.001). (C) The panel below the graph shows SEM images of mixed-species biofilms after 72 h of incubation in the combinations indicated.

A different picture emerged in the case of the mixed GAS/S. oralis biofilms. Both short-chain growing S. oralis strains predominated the bottom layer and long-chain growing GAS strains in the top layer of the two-species biofilm (Fig. 4; see also Fig. S3 in the supplemental material), indicating that S. oralis acted as a support for GAS biofilm development. Obviously, the biofilm was not as exclusively formed by S. oralis, as was the case for S. salivarius, but this time was somewhat affected by the biofilm-forming capacity of the individual GAS strains. In the case of GAS M49, biofilm masses in combination with the S. oralis strains were significantly higher than the biofilm masses of GAS M49 or each S. oralis strain alone (Fig. 4A). In contrast, biofilm masses of GAS M6 combined with either S. oralis strain did not significantly differ from that of single-species biofilms (Fig. 4B).

Artificial saliva effects on biofilms.

Copious amounts of saliva are present in parts of the human upper respiratory tract. Besides physiological effects on host surfaces, it has an important role in microbial homoeostasis on these surfaces (44). Therefore, the effect of artificial saliva was determined for monospecies and mixed-species biofilms.

Artificial saliva reduced all monospecies biofilm masses except for S. oralis DSMZ (see Fig. S4 in the supplemental material). The addition of artificial saliva also reduced the masses of the mixed-species biofilms involving S. salivarius (Fig. 5). The predominance of S. salivarius in combinations with GAS disappeared upon saliva addition. Now, both GAS strains prevailed in the SEM pictures, the M6 strain more than the M49 strain (compare Fig. 4 to 5).

Fig 5.

Impact of artificial saliva on biofilm formation of cocultures of S. pyogenes M49 (A) or M6 (B) and respiratory tract streptococci. After 72 h of incubation, the biofilms were quantified by safranin staining and absorption measurements at 492 nm. Light gray bars represent mixed-species biofilms in the presence of artificial saliva, and dark gray bars represent mixed-species biofilms in the absence of artificial saliva. Error bars resulted from three independent experiments, asterisks indicate significant differences (Student t test: ***, P < 0.001). (C) The panel below the graph shows SEM images of the mixed-species biofilms after 72 h of incubation in the presence of artificial saliva.

The results for the GAS/S. oralis combinations in the presence of artificial saliva diverged in a strain-dependent manner. Combinations of the two GAS strains with S. oralis DSMZ did not lead to a statistically significant change in biofilm masses (Fig. 5). The effect of artificial saliva on the GAS/S. oralis 4087 combinations differed for the M49 and M6 strains. In combination with the M49 strain, the biofilm mass was drastically decreased, whereas the biofilm mass was unaffected by saliva in the S. oralis 4087/GAS M6 combination. Consistently, artificial saliva did not change the two streptococcal biofilm layers formed by the S. oralis and GAS strains (Fig. 5).

Effects of respiratory tract streptococci on GAS host cell adherence and internalization.

After obtaining data on the mixed-species biofilm behavior of GAS and respiratory tract streptococci on inanimate supports, we were now interested in this interaction on host cell surfaces, which are typical targets for GAS adherence and internalization. Different experimental setups were chosen to reflect all possible interaction scenarios: (i) GAS strains were first allowed to contact the HEp-2 cells prior to adding the other respiratory tract streptococci; (ii) the other respiratory tract streptococci were initially seeded to the HEp-2 cells, and only thereafter were GAS introduced into this setting; and (iii) both species were simultaneously added to the host cells. With each strategy, we examined whether the respiratory tract streptococci could increase, delay, or even cure harmful effects inflicted by GAS on host cells.

Compared to consequences after single-species exposure, only strategy ii had significant effects. The initial presence of the respiratory tract streptococci led to a marked reduction in HEp-2 cell adherence and internalization for both GAS serotype strains (Fig. 6). Consistently, cytotoxicity of GAS toward host cells was reduced in this setting (see Fig. S5 in the supplemental material). In conclusion, S. salivarius and both S. oralis strains protected HEp-2 cells from the attack of both GAS strains to a comparable degree; however, this only happened when the respiratory tract streptococci first interacted with the host cells.

Fig 6.

Impact of respiratory tract streptococci on adherence and internalization of S. pyogenes to/into HEp-2 cells. The numbers of adherent (dark gray bars) and internalized (light gray bars) bacteria of S. pyogenes M49 (A) and M6 (B) alone on HEp-2 cells were set to 100%. The adherence and internalization values for both S. pyogenes strains after pretreatment of the HEp-2 cells with the respiratory tract streptococci were related to the values in the absence of the respiratory tract streptococci. Error bars reflect data from three independent experiments. (C) The adherence of both GAS strains (green) to HEp-2 cells (gray) is shown in a fluorescence/light microscopy overlay picture.

Modifications of the experimental setup, such as a washing step between the application of the different bacteria and a transwell setup with respiratory tract streptococci in the upper compartment and GAS plus host cells in the lower compartment, were introduced in order to uncover details of the protection mechanism. As a result of both modifications, an unchanged presence of the respiratory tract streptococci in the immediate vicinity of the host cells is crucial for the observed protective effects (see, for example, Fig. S6 in the supplemental material) for the adherence and internalization of the M49 strain.

SEM pictures of the interaction of the respiratory tract streptococci with the HEp-2 cells (Fig. 7) demonstrated the presence of high numbers of S. salivarius K-12 bacteria forming biofilm-like structures on the host cells. Both S. oralis strains were present in only low numbers on the host cell surfaces and did not attain biofilm-like structures. This finding suggested that the similar protective effects exerted by the respiratory tract streptococci could be based on different functional or molecular backgrounds. The S. salivarius strain could potentially act as steric barrier, whereas the S. oralis strains could use molecular ways to influence GAS-host cell interaction.

Fig 7.

Adherence of respiratory tract streptococci to HEp-2 cells. The SEM images show the adherence of S. salivarius K-12 (A), S. oralis 4087 (B), and S. oralis DSMZ (C) strains to HEp-2 cells.

Transcriptional HEp-2 cell responses upon exposure to streptococcal strains.

In order to clarify some of the molecular details of the different mechanisms that the respiratory tract streptococci used for their similar protection of HEp-2 cells, we examined the host cell transcriptomes after exposure to the protective streptococci by utilizing Affymetrix whole-genome transcriptome microarrays. HEp-2 cells were incubated with the M49 GAS strain for 2 h as positive controls. In parallel, HEp-2 cells were preincubated for 2 h with respiratory tract streptococci, and then GAS M49 was added, followed by incubation for another 2 h. In subsequent data analysis, at least 3-fold altered transcript amounts (signal log ratio ≥ 1.58) were defined as a threshold for significance.

Pre-exposure of HEp-2 cells to S. oralis DSMZ revealed 10 genes with annotated functions that were differentially expressed compared to cells exposed to GAS M49 alone. Of these genes, only one (endothelin-2 gene EDN2) displayed an increased transcript amount, while transcripts of nine genes were present at a lower level (see Table S1 in the supplemental material). Among the genes with decreased transcript amounts was PLG (plasminogen), which has been associated with bacterial adhesion and internalization (45).

Similarly, analysis of the HEp-2 cell transcriptomes after preincubation with S. salivarius and subsequent GAS M49 exposure compared to GAS M49 exposure alone revealed 19 annotated genes with an at least 3-fold difference in transcript amounts. Seventeen of these genes exhibited lower transcript amounts, whereas only two genes showed higher transcript amounts (see Table S2 in the supplemental material).

DISCUSSION

S. pyogenes as an exclusive human pathogen uses various routes to enter the host. Uptake via airborne droplets is the predominant way of entering the respiratory tract and causing infections such as pharyngitis and tonsillitis. Many details of the first infection steps at this anatomic site have been elucidated in the last several decades concerning the involved GAS virulence factors and mechanisms of pathogen-host cell interactions (3, 7, 9). However, before attaching to host tissues GAS will encounter the resident microbial biome. The published information about that incident is still scarce. Thus, the present work had several major objectives.

First, the consequences of mixing different GAS serotype strains with diverse respiratory tract streptococci should be examined and quantified in planktonic and biofilm states under aspects of survival and virulence factor expression. Second, mixed-species biofilm development with the different species should be scrutinized since the specific GAS target epithelia on the pharynx and tonsils are especially covered by bacterial biofilms (46). Third, the effect of respiratory tract streptococci on the interaction between GAS and host cells should be inspected with specific regard to an antecedent, simultaneous, or subsequent presence of the potentially protective streptococci. Since the findings could have therapeutic implications, besides representatives of the leading local species, i.e., S. oralis (47, 48) and an established probiotic bacterium, the S. salivarius K-12 strain, were chosen as interaction partners.

Streptococci are the predominant bacteria in the oral cavity and are present in numbers ranging between 5 and 8 logs per ml of fluid or mg of material (46). Much less is known about the numbers of S. pyogenes bacteria when introduced into the pharynx. Most probably, S. pyogenes will not be taken up as single bacteria but inhaled in droplets or ingested as dried material containing densely packed bacteria that could equal several logs. To assess numerical changes as a consequence of simultaneous presence of potential antagonists, S. pyogenes wild-type strains were mixed with respiratory tract bacteria in various ratios to reflect the numerical variability of the natural environment. High numbers of planktonic S. salivarius and S. oralis generally suppressed the growth of S. pyogenes—more pronounced for the M49 than for the M6 serotype strain—in direct contact experiments. However, in transwell assays separating both growth partners, as well as in classical bacteriocin assays on solid media, only S. salivarius displayed this effect on the growth of both GAS serotype strains, implying the production of diffusible active substances specifically by this species. The latter result was consistent with recent findings on this probiotic bacterium (49, 50), provided the S. pyogenes strain was susceptible to the substance(s) and the cell numbers of the probiotic strain exceeded that of S. pyogenes by several logs. The present results extend the published data since thus far growth inhibition testing of S. pyogenes has only been conducted on solid media or in liquid assays using (semi)purified bacteriocin. Several bacteriocin-producing S. salivarius strains were shown to have probiotic effects and inhibit Gram-positive pathogens of the upper respiratory tract such as S. pneumoniae or S. pyogenes (51), e.g., by the production of the lantibiotic salivaricin D (52). However, the S. salivarius-produced salivaricin could induce the production of defensive agents such as the S. pyogenes bacteriocin SalA1. In the present study, the growth of the respiratory tract streptococci was not suppressed in the presence of diffusible substances from the GAS strains. In in situ studies, S. pyogenes SalA1 expression was induced by a minimum of 8 × 10⁵ S. salivarius cells per ml of saliva (49). This number is above the levels of S. salivarius in saliva from healthy children, i.e., 10⁵ CFU/ml (53) and 10⁴ to 10⁵ CFU/ml of saliva from adults treated with the corresponding probiotic strains (54). Therefore, a GAS-associated growth inhibition of respiratory tract streptococci could be irrelevant for the natural encounter between both bacterial species in their human hosts.

Nutrient competition-derived reduced growth of one species in the presence of large inocula of another species could be the explanation for the observed effects of planktonic S. oralis on GAS and vice versa. However, the strain specificity of this effect for S. pyogenes, as well as the exclusive occurrence in direct contact but not in transwell system experiments, insinuated the presence of additional mechanisms. Such cell-associated but still undefined factors have also been observed in cocultures between S. oralis/Haemophilus influenzae or Moraxella catarrhalis (55) or cocultures between E. faecalis/S. pneumoniae, S. aureus, or Listeria ivanovii (56, 57).

S. pyogenes has been demonstrated to form single-species biofilms in vitro and in animal model infections (29, 32, 58). The protective function of a biofilm leading to an increased antibiotic resistance level stimulated the idea that biofilms could be associated with long-term persistence in asymptomatic carrier persons (59). However, in the pharynx, S. pyogenes encounters preexisting mixed-species biofilms on its target cells. Thus, it has to establish itself within and eventually to penetrate these biofilms. Based on data from other pathogens, it is conceivable that both establishment and penetration will be influenced by many factors produced by the resident microbial biome (60–63).

Consequently, respiratory tract streptococci were used to study defined mixed-species biofilms with the serotype M49 and M6 S. pyogenes strains. Here, we could show that in general S. pyogenes can form mixed-species biofilms with the streptococcal species under investigation. In most cases, increased biofilm masses were detected. Of note, GAS M49, which forms scanty biofilms as single species, was stimulated to a higher production rate in the presence of S. oralis. Moreover, in all combinations with S. oralis strains, S. oralis formed the bottom layer and S. pyogenes formed the top layer (Fig. 4). This is consistent with the pioneer character of S. oralis also observed in ex vivo samples from human volunteers (64) and in models of oral biofilms (65). The avidity of S. oralis for human cell surfaces seems to be associated with the activity of specific galactose- or sialic acid/N-acetylgalactosamine- binding lectins (66, 67). Also, the interaction between S. oralis and S. pyogenes could rely on the activity of lectins and appropriate sugar moieties, since the latter species was shown to bind the lectin concanavalin A (33), as well as several human glycoproteins (68).

Since pharyngeal biofilms would be bathed in saliva, the experiments were repeated in the presence of mucin as the major component of saliva (44). Generally, S. pyogenes should be able to deal well with saliva. It was shown to bind to mucin by its M protein (69). Simultaneously, the two-component regulator SptRS is induced by the presence of saliva (70). This control circuit induces virulence factors such as Sic and SpeB, which inactivate antimicrobial peptides normally contained in saliva (70–73). Other saliva-driven regulatory circuits (MalR and CcpA) couple the usage of maltodextrin, the predominant sugar in saliva, as an exclusive energy source with the expression of several virulence factors and thereby support its respiratory tract persistence (74, 75). However, in earlier publications it was shown that saliva decreased the ability of S. pyogenes to attach to eukaryotic epithelial cells (76).

In testing single-species biofilms, S. pyogenes and S. salivarius biofilm masses were strongly diminished by artificial saliva. Artificial saliva has been described to facilitate S. salivarius growth but counteracts its viability during stationary phase and under starvation conditions (77).

In mixed-species biofilms, biofilm masses were minimized for blends of both GAS isolates with S. salivarius, whereas in combinations with the S. oralis DSMZ strain, artificial saliva had comparatively small or rather augmenting effects. This augmenting effect of artificial saliva on biofilm formation of both S. oralis DSMZ single- and mixed-species biofilms is consistent with the species' capacity to use mucin or the mucin component sialic acid as a carbon source by expressing secreted enzymes such as β-d-galactosidase, β-d-glucosidase, and β-N-acetyl-d-glucosaminidase (78–80). However, we observed the contrary in the S. oralis 4087 strain, on which saliva apparently exerts a negative influence on its biofilm-forming capacity. Such strain-specific effects of saliva on S. oralis have recently been described and putatively been associated with the expression of specific adhesins (81).

Once integrated and established in the biofilms, GAS is located closer to the underlying epithelial cell layers. Binding of S. pyogenes to eukaryotic cells is indispensable for causing disease and persisting in its human host (3). The influence of simultaneously present bacteria on this interaction is not well studied. Thus far, only for Moraxella catarrhalis, another nasopharyngeal pathogen, was coaggregation with S. pyogenes reported to lead to increased eukaryotic cell adherence but decreased internalization (82, 83).

To collect more data on the interaction of two bacterial and one eukaryotic partner, HEp-2 respiratory epithelial cells were exposed to combinations of the two S. pyogenes strains and the S. oralis and S. salivarius strains. When testing the growth of single species on the eukaryotic cells by SEM inspection, S. pyogenes strains formed structures resembling microcolonies, S. salivarius formed large masses on the cell surface, while the S. oralis strains only occasionally adhered to the eukaryotic cells as single cells or pairs (Fig. 7). Although the molecular basis for S. pyogenes eukaryotic cell attachment and microcolony formation is well established (28, 31, 33), for S. salivarius thus far only surface antigen C has been determined as a responsible adhesin (84, 85).

S. salivarius and S. oralis strongly reduced S. pyogenes HEp-2 cell adhesion and internalization if present for 2 h before S. pyogenes was seeded (Fig. 6). Obviously, the established probiotic S. salivarius K-12 strain and the two S. oralis isolates could efficiently protect eukaryotic cells from S. pyogenes adherence, provided the streptococcal strains have sufficient time to establish themselves before S. pyogenes is inoculated. According to the results of the bacteriocin assays (Fig. 2), this protection was not associated to decreased GAS hemolysin production. Since S. salivarius was demonstrated by microscopy and quantitative measurements to form biofilm-like structures on eukaryotic cells, while S. oralis hardly bound to this target, the protection is most probably based on different mechanisms.

In addition to the killing of planktonic GAS (Fig. 1), the eukaryotic cell protection exerted by S. salivarius could be due to steric effects, i.e., the biofilm-like structure by which these bacteria cover the cells. Such protective effects of S. salivarius K-12 have also been demonstrated against candidiasis in mouse models after oral application (86). For the S. oralis strains, which, according to the results from the cytotoxicity assays (see Fig. S5 in the supplemental material), guard the HEp-2 cells even more efficiently, only small bacterial amounts bind to host cells, thereby precluding the steric hindrance hypothesis. Thus, most probably another mechanism applies. Potentially, regulatory and metabolic pathways in the eukaryotic cells could be influenced by the presence of the S. oralis strains rendering the cells less susceptible to the S. pyogenes attack. Such beneficial effects on eukaryotic cells, as well as the opposite, have been described for several viridans streptococci (87–89). In contrast to the findings of Cosseau et al. (87) with the human bronchial epithelial cell line 16HBE14O, a notably few number of transcripts showed significantly altered abundances in HEp-2 cells first exposed to respiratory tract streptococci and then to M49 GAS compared to the positive control: 19 in the case of S. salivarius and 10 in the case of S. oralis. Among the genes repressed in the presence of S. oralis is PLG, encoding for plasminogen, which has been shown to be involved in the adherence and internalization of GAS M49 to or into host cells (45). Although the production of an active compound has been described in S. salivarius that inhibits the NF-κB pathway in human epithelial cells and reduces secretion of the proinflammatory interleukin-8 (IL-8) (90), IL-8 transcript amounts were unaltered in our experimental setup. Both streptococcal species lead to a downregulation of genes involved in cGMP, GTPase, RAB, and RAS signaling (GFM1, HRSALS, GUCA1B, RAB6A, and RAPGEF2), as well as a changed expression of genes involved in apoptosis regulation (XRCC5, CCAR1, and MNDA). Furthermore, upregulation of the expression of endothelin-2 in the case of S. oralis could lead to an enhanced activation and acquisition of neutrophils and macrophages in vivo.

Taken together, the study showed that S. oralis could induce protection of the eukaryotic cells even without largely binding to the cells or producing bacteriocins affecting S. pyogenes. Also, transcriptome analysis does not uncover apparent mechanisms by which S. oralis could protect the eukaryotic cells. The surprisingly less effective protection of HEp-2 cells by S. salivarius also involved expression changes in only a restricted panel of genes and, instead, could predominantly be exerted by building an almost impermeable, potentially bacteriocin-producing wall of S. salivarius biofilm in front of the host cell target.

We demonstrated here that S. pyogenes can establish itself in mixed-species biofilms with typical species of the resident pharyngeal microbial biome. However, the species, i.e., S. oralis, which most efficiently supports S. pyogenes growth in mixed-species biofilms and simultaneously does not affect virulence factor production in the beta-hemolytic streptococci, also most efficiently protects underlying epithelial host cells from S. pyogenes-inflicted damages.