Aggregatibacter actinomycetemcomitans mediates protection of Porphyromonas gingivalis from Streptococcus sanguinis hydrogen peroxide production in multi-species biofilms

Mixed species biofilms are shaped and influenced by interactions between species. In the oral cavity, dysbiosis of the microbiome leads to diseases such as periodontitis. Porphyromonas gingivalis is a keystone pathogen of periodontitis. In this study, we showed that polymicrobial biofilm formation promoted the tolerance of Porphyromonas gingivalis to oxidative stress under micro-aerobic conditions. The presence of Streptococcus sanguinis, an oral commensal bacterium, inhibited the survival of P. gingivalis in dual-species biofilms via the secretion of hydrogen peroxide (H2O2). Interestingly, this repression could be attenuated by the presence of Aggregatibacter actinomycetemcomitans in tri-species biofilms. It was also shown that the katA gene, encoding a cytoplasmic catalase in A. actinomycetemcomitans, was responsible for the reduction of H2O2 produced by S. sanguinis, which consequently increased the biomass of P. gingivalis in tri-species biofilms. Collectively, these findings reveal that polymicrobial interactions play important roles in shaping bacterial community in biofilm. The existence of catalase producers may support the colonization of pathogens vulnerable to H2O2, in the oral cavity. The catalase may be a potential drug target to aid in the prevention of periodontitis.

distributed in the oral cavity 9,10 . We performed experiments to evaluate the survival of P. gingivalis in 3 environments: 14 mL test tubes with shaking at 100 rpm under micro-aerobic conditions (6% oxygen, gas mixture), and 4-well chambers without shaking (static) under either anaerobic (0% oxygen, gas mixture) conditions or micro-aerobic conditions. The incubator shaking was used to inhibit biofilm growth and thus encourage planktonic growth while the lack of shaking was used to facilitate biofilm formation. All three environments had an initial inoculation of 1 × 10 8 P. gingivalis ATCC 33277 (Pg) cells into CDM and were incubated for four days at 37 °C. When Pg was cultured in the 14 mL tubes environment, it was not able to survive (Fig. 1A). Pg survived in both 4-well chamber environments; the biomass under micro-aerobic conditions was significantly lower than that under anaerobic conditions (P ≤ 0.01) (Fig. S1). Pg cells from the micro-aerobic 4-well chamber were then inoculated onto agar plates and grown in order to verify that the detected fluorescent signal was from live cells. The growth of 2.03 × 10 6 ± 9.71 × 10 5 colony-forming units (CFUs) confirmed the existence of live biofilm cells in static micro-aerobic conditions (Fig. 1A). Although the CFU of Pg grown under micro-aerobic conditions in 4-well chambers was lower than that of the initial inoculation, it was still greater than the CFU of Pg grown in the 14 mL test tube, suggesting that the biofilm formation increased the tolerance of Pg to oxidative stress from the presence of environmental oxygen (Fig. 1A). Because there have already been a number of reports illustrating that biofilm formation increase the tolerance of bacteria to oxidative stress 33 , the next experiment was designed to focus on the effect of mixed-species biofilm bacterial interactions on the oxidative stress tolerance of P. gingivalis.
To examine Pg survival in multi-species biofilms, four groups of bacterial mixes were tested: Pg only, Pg and A. actinomycetemcomitans 652 (Aa), Pg and S. sanguinis SK36 (Ss) and a mixture of all three species. Each bacterial mix group was incubated in CDM medium under micro-aerobic conditions in 4-well chambers. Biofilms were first grown for four days, after which they were stained by fluorescence in situ hybridization (FISH) and were visualized using confocal laser scanning microscopy (CLSM). Biofilm biomass was quantified by COMSTAT script in Matlab software 33 .
All four groups of biofilms were successfully detected using FISH probes. Under these micro-aerobic conditions, the biomass of Pg in biofilm was not significantly changed by the presence of Aa when compared to the Pg single species biofilm control (Fig. 1B,C). In contrast, the presence of Ss significantly lowered the biomass of Pg in Ss-Pg dual-species biofilms, suggesting that Ss was dominant and somehow inhibited survival of Pg (0.442 ± 0.083 µm 3 /µm 2 in Pg only biofilm and 0.021 ± 0.009 µm 3 /µm 2 in Pg-Ss dual species biofilm (P ≤ 0.001)) ( Fig. 1B,C). Interestingly, the biomass of Pg in Pg-Aa-Ss tri-species biofilms was significantly increased compared to that in Pg-Ss dual-species biofilms, implying that Pg survival inhibition by Ss could be partially attenuated by the presence of Aa (Fig. 1B,C). As there was no significant difference between the biomass of Pg from Pg single species www.nature.com/scientificreports www.nature.com/scientificreports/ biofilms and Pg-Aa dual-species biofilms, it is feasible to suggest that Aa could interact with Ss, counteracting the influence of Ss on Pg, indirectly promoting the survival of Pg.
Ss-produced H 2 o 2 reduced the biomass of Pg. H 2 O 2 is a well-studied inhibitory mechanism that S. sanguinis uses to compete with Streptococcus mutans 19,34 . It can be generated by a pyruvate oxidase (SpxB) in Ss via a reaction converting pyruvate to acetyl phosphate. During this catalytic process, oxygen is consumed 35,36 . Pg-Ss dual-species biofilms were grown under micro-aerobic conditions. We tested the inhibitory ability of Ss-produced H 2 O 2 by decomposing H 2 O 2 with 10,000 U/mL of catalase (Catalase from bovine liver, Sigma). More Pg was present in the dual-species biofilms when catalase was supplemented in the medium (P ≤ 0.001) ( Fig. 2A). This result suggested that H 2 O 2 was essential for Ss to inhibit the survival of Pg under micro-aerobic conditions. Pg appeared to preferentially colocalize with Ss in Pg-Ss dual-species biofilms when catalase was supplemented. This phenomenon implied that Pg might coaggregate with Ss in multi-species biofilms, which was similar to the interaction between P. gingivalis and S. gordonii 29 .
When Pg was co-cultured with Ss ΔspxB, the biomass of Pg was still greatly inhibited by Ss ΔspxB (Fig. 2B). One probability was that, in contrast with Ss wild type (WT), the ΔspxB mutant could still produce about 25% the concentration of H 2 O 2 19 , which might be enough for the inhibition of Pg growth. Since more Pg survived in the Pg-Aa-Ss tri-species biofilm, the supplementation of Aa might have a better effect than the deletion of spxB on reducing H 2 O 2 concentration. To test this hypothesis, Ss WT and Ss ΔspxB biofilms with/without the addition of Aa were cultured. The amount of H 2 O 2 in the supernatant of the 4-day old biofilms was measured using Hydrogen Peroxide Assay. Indeed, the H 2 O 2 concentration in the Aa-Ss WT dual-species biofilm was much lower than that in the Ss ΔspxB single species biofilm (P ≤ 0.0001), which supported the hypothesis (Fig. 3A). Additionally, in comparison to the Ss WT single species biofilm, the Ss ΔspxB single species biofilm contained less H 2 O 2 in the supernatant (P ≤ 0.0001), which was consistent with the result in the previous study showing that the spxB gene deletion decreased H 2 O 2 production in Ss (Fig. 3A) 19 .
The Aa-Ss ΔspxB dual-species biofilm contained less H 2 O 2 than the Aa-Ss WT biofilm, which indicated that the spxB gene deletion might promote Pg survival in Pg-Aa-Ss tri-species biofilms (Fig. 3A). 4-day old Pg-Aa-Ss WT and Pg-Aa-Ss ΔspxB tri-species biofilms were treated with FISH and observed by CLSM. The biomass of Pg in the Pg-Aa-Ss ΔspxB tri-species biofilm was more than that in the Pg-Aa-Ss WT biofilm (P ≤ 0.01), which suggested that the H 2 O 2 produced by Ss played an important role in inhibiting Pg growth (Fig. 3B,C).
Similar to Pg, the biomass of Aa was also increased in the Pg-Aa-Ss ΔspxB biofilm (P ≤ 0.001) (Fig. 3B,C). Surprisingly, the biofilm biomass of Ss was increased in the Pg-Aa-Ss ΔspxB biofilms than that in the Pg-Aa-Ss WT tri-species biofilm (P ≤ 0.05) (Fig. 3B,C), despite Ss ΔspxB had a reduced biofilm formation in Ss-Pg dual-species biofilms (Figs 2B and S2). When we treated tri-species biofilms using FISH protocol, we observed that the Pg-Aa-Ss WT biofilm was more fragile than the Pg-Aa-Ss ΔspxB biofilm, indicating that H 2 O 2 might affect inter-species attachment. It has been shown that P. gingivalis utilizes fimbrillin to bind to glyceraldehyde-3-phosphate dehydrogenase, a cell surface protein of S. sanguinis 37 , implying that S. sanguinis may not only inhibit the growth of P. gingivalis but also coaggregate with P. gingivalis. Due to the deletion of spxB, a reduced antagonism in Pg-Aa-Ss tri-species biofilm might be beneficial for the co-aggregation between Pg and www.nature.com/scientificreports www.nature.com/scientificreports/ Ss, and as a result, might increase the biomass of Ss. Though a similar relationship may exist between S. sanguinis and A. actinomycetemcomitans, the current knowledge on their interactions is limited, and the mechanism of such phenomenon needs further exploration.  (Fig. 4A).

Aa degraded H 2 o 2 and protected
Subsequently, the H 2 O 2 produced by Ss, Pg and Aa was tested. Compared with the blank control (CDM without bacteria), neither Pg nor Aa produced H 2 O 2 (Fig. 4B). The H 2 O 2 concentration in Aa was even lower than the concentration in the blank control (Fig. 4B). Ss produced almost 0.3 μM of H 2 O 2 after 30 minutes of reaction, which could be attenuated by the addition of Aa (P ≤ 0.001) but not Pg (Fig. 4B). The presence of Aa decreased the H 2 O 2 concentration by nearly half in both the dual-species (Ss-Aa) and tri-species (SS-Pg-Aa) suspensions (Fig. 4B). There was no significant change in cell density during the 30 minutes of experimentation, indicating that cell growth did not influence the results of H 2 O 2 concentrations (Fig. S3). Similar results were reported in previous works, which utilized scanning electrochemical microscopy to do real-time mapping of H 2 O 2 concentrations on bacteria biofilms 38 . They reported that the H 2 O 2 generated by S. gordonii, another oral commensal www.nature.com/scientificreports www.nature.com/scientificreports/ bacterium and H 2 O 2 producer, could be reduced by A. actinomycetemcomitans 38 . These data suggested that Aa degraded H 2 O 2 produced by Ss and implied that Aa might be able to promote the survival of Pg in Pg-Aa-Ss tri-species biofilm by reducing H 2 O 2 concentration.
Bacteria living in biofilms are surrounded by matrix composed of polysaccharide, eDNA and proteins 39 . As materials may slow penetrate and transverse a biofilm 39 , cell to cell distance may impact the interaction between Aa and Ss. To test the contribution of cell-cell distance to the Aa-Ss interaction, Aa and Ss were cultured in a transwell system, where they were separated by a 0.4 μm filter. Ss was cultured at the bottom and Aa was either incubated in the insert or mixed with Ss at the bottom. After 30 minutes of reaction, the H 2 O 2 concentration at the bottom of the well was measured. The H 2 O 2 concentration at the bottom was 2.404 ± 0.035 μM when the insert was filled with CDM medium and the bottom was Ss. When Aa was put in the insert and Ss was set at the bottom of the well, Aa slightly but significantly decreased the H 2 O 2 concentration at the bottom to 2.087 ± 0.061 μM (P ≤ 0.05) (Fig. 4C). However, the reduction was much lower than that in the well where Ss and Aa mixed directly at the bottom (P ≤ 0.001) (Fig. 4C). This result showed that a closer distance between Ss and Aa was beneficial for Aa to reduce H 2 O 2 produced by Ss. Aa might have limited function to degrade H 2 O 2 when it was far away from Ss. In an in vitro study, Aggregatibacter has been shown to close contact with Streptococcus 40 , indicating that A. actinomycetemcomitans might exist near to S. sanguinis in vivo to detoxify H 2 O 2 . www.nature.com/scientificreports www.nature.com/scientificreports/ KatA has been reported to produce catalase in A. actinomycetemcomitans strain VT1169 (Aa VT1169) to detoxify H 2 O 2 and is essential for the survival of A. actinomycetemcomitans during co-infection with S. gordonii 24,25 . It was hypothesized that KatA was essential for Aa to improve the survival of Pg.
Using the Hydrogen Peroxide Assay Kit, H 2 O 2 concentration and cell density of Pg + Ss + Aa VT1169 and Pg + Ss + A. actinomycetemcomitans VT1169 ΔkatA (Aa ΔkatA) suspensions were monitored. Compared to the katA deletion mutant, Aa VT1169 had the greater ability to repress H 2 O 2 production (P ≤ 0.001 at the time point of 110 minutes), implying that KatA was important for Aa to reduce the H 2 O 2 generated by Ss (Fig. 5A). There was no significant difference in cell density between Pg + Ss + Aa VT1169 and Pg + Ss + Aa ΔkatA, suggesting that the difference in H 2 O 2 concentration was not caused by a difference in cell growth (Fig. S4).
Pg-Ss-Aa VT1169 and Pg-Ss-Aa ΔkatA Tri-species biofilms were stained by FISH and observed by CLSM as described above. Compared with the biofilm of Pg-Ss-Aa VT1169, the Pg-Ss-Aa ΔkatA biofilm contained less Pg and Aa (P ≤ 0.001 for both comparisons) (Fig. 5B,C), which suggested that the catalase of A. actinomycetemcomitans was essential for the survival of both Pg and A. actinomycetemcomitans in the tri-species biofilms and further confirmed the hypothesis that A. actinomycetemcomitans protected P. gingivalis from H 2 O 2 damage. The biofilm biomass of Ss in Pg-Ss-Aa ΔkatA was slightly decreased (P ≤ 0.05) (Fig. 5B,C). This phenomenon where Ss biofilm biomass decreased in conditions that also led to decreased biofilm biomass of Aa and Pg, was similar to the observed phenomenon in Fig. 3B,C allowing for the possible hypothesis that the biomass of Ss might have been impacted by Pg and/or A. actinomycetemcomitans in tri-species biofilms.
The VT1169 and ΔkatA single species biofilms were stained by SYTO9 and observed by CLSM. The morphology of ΔkatA biofilm was different from that of the wild type strain. The biofilm of ΔkatA was much thicker (Fig. S5A). It contained larger aggregations and bigger gaps between aggregations (Fig. S5A). However, the biofilm biomass of these two strains were similar, which confirmed that the reduction of Aa biomass in Pg-Ss-Aa ΔkatA tri-species was not caused by an attenuated biofilm formation ability of Aa ΔkatA (Fig. S5).  www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
In this study, it was shown that Aa degraded H 2 O 2 produced by Ss, which consequently aided the survival of Pg in Pg-Aa-Ss tri-species biofilms under micro-aerobic conditions. KatA, which produces catalase in Aa, was also shown to participate in this interaction. There have been many epidemiological studies showing that anaerobic bacteria such as P. gingivalis exist in supragingival, salivary and mucosal samples 9,10 . One possibility is that micro-environment may exist in oral cavity, which allows these anaerobic bacteria to survive in micro-aerobic conditions. The results of this study presented the possibility that catalase producers in oral microbiota attenuate the oxidative stress and help the survival of anaerobic species under micro-aerobic conditions.
The pathogenesis of periodontitis has been thought to possess polymicrobial synergistic interactions 41,42 . A temporal dynamics study showed that facultative anaerobic bacteria, especially Streptococcus, were dominant in the early stage of oral biofilm formation 43 . Subsequently, the 'healthy' biofilm composition was replaced with a population of gram-negative anaerobic bacteria 43 . In our study, we observed essentially no Pg presence in Pg-Ss dual species biofilms but Pg presence was obvious in Pg-Aa-Ss tri-species biofilms, suggesting that the existence of Aa was important for Pg survival. This result indicated that the earlier colonization of bacteria species with catalase activity than anaerobic species in oral biofilms might be necessary to generate suitable surroundings for the survival of anaerobic microorganisms. Further studies need to be performed to test this hypothesis. Additionally, our study illustrated that KatA of Aa VT1169 was important for the growth of both Pg and Aa VT1169, implying that catalase might be a promising drug target to prevent periodontitis.
Welch et al. utilized FISH technology to stain supragingival dental plaque 40 . They hypothesized that the Porphyromonas growing at the periphery of biofilm samples might not be P. gingivalis because the outer shell of the biofilm was in a presumably aerobic environment 40 . Here, we demonstrated that P. gingivalis was able to survive in a micro-aerobic environment and had better survival in the presence of A. actinomycetemcomitans, which implied that it was possible that the bacteria at the periphery of supragingival biofilm samples, seen in the Welch's study, was P. gingivalis. In their study, they showed that both Porphyromonas and Haemophilus/Aggregatibacter were in close contact with Streptococcus cells 40 . Furthermore, Aggregatibacter was not found adjacent to cells of Porphyromonas in the absence of Streptococcus 40 . Their results indicated that P. gingivalis, A. actinomycetemcomitans and S. sanguinis might be close to each other in vivo and a similar interaction between these three species might also exist in vivo.
In Fig. 2A, Pg appeared to preferentially colocalize with Ss in Pg-Ss dual-species biofilms when catalase was supplemented. Additionally, the biomass of Ss in both Figs 3B and 5B were positively related with the biomass of Pg and Aa. All the phenomena above indicated that Ss might also cooperate with Pg and/or Aa in multi-species biofilms. The antagonism and the cooperation between commensal bacteria and pathogens may exist in equilibrium in oral microbiota. Whenever the antagonism was weakened, or the cooperation was strengthened either by other microorganisms or environmental conditions, dysbiosis may happen and lead to diseases such as periodontitis.

Materials and Methods
Bacterial strains, growth and antibiotics. Strains used in this study are listed in Table S1. Unless otherwise stated, Pg, A. actinomycetemcomitans strains and Ss cells from −80 °C frozen glycerol stocks were 0.5% inoculated into TSB medium (tryptic soy broth supplemented with yeast extract (5 mg/ml), hemin (5 µg/ml) and menadione (1 µg/ml)) and incubated statically under anaerobic conditions (10% CO 2 , 10% H 2 and 80% N 2 ) at 37 °C using an Anoxomat ® system (Spiral Biotech, Norwood, MA). Spectinomycin was used at 50 μg/mL for the culture of Aa ΔkatA. No antibiotic was added to multi-species biofilms. The CFUs of Pg were tested by growing Pg on sheep blood agar plates (Trypticase ™ Soy Agar (TSA II ™ ) with Sheep Blood, BD BBL ™ ) under anaerobic conditions. All media was incubated in anaerobic jars for at least 2 days before experiments to equilibrate. Biofilm assay. Pg, A. actinomycetemcomitans strains and Ss were initially incubated separately for 48 hours, 24 hours and overnight respectively, in TSB medium under anaerobic conditions to early stationary phase. The resultant growth was then resuspended in fresh CDM, followed by 10% inoculation into CDM medium and incubation under micro-aerobic conditions for biofilm formation. CDM was prepared as previously described  44 . Biofilms were incubated in 4-chambered glass coverslip wells (Chambered Coverglass, Nunc ™ Lab-Tek ™ ) for 4 days at 37 °C. Cultures were grown anaerobically (0% O 2 , 10% CO 2 , 10% H 2 and 80% N 2 ) or micro-aerobically (6% O 2 , 7.2% CO 2 , 7.2% H 2 and 79.6% N 2 ) in jars using the Anoxomat ® system (Spiral Biotech, Norwood, MA). Single-species biofilms were stained using SYTO 9 (SYTO ™ 9 Green Fluorescent Nucleic Acid Stain, Invitrogen ™ Molecular Probes ™ ) and FISH was used to analyze and characterize the composition in multi-species biofilms. FISH assay. FISH was performed as previously described 40 . FISH probes used in the study were ordered from Integrated DNA Technologies, Inc. and the sequences were listed in Table S2. Biofilms were grown in 4-well chambers for 4 days in 1 mL of CDM medium. 800 µL of supernatant was discarded by pipetting and then 4-well chambers were slowly turned over on paper towels to discard remaining supernatant. Biofilms were gently washed by 200 µL of 1× PBS buffer and fixed by 2% (wt/vol) paraformaldehyde on ice for at least 1.5 hours. After fixation, samples were gently washed again in 1× PBS for 15 min. Next, PBS was discarded and 10 µL of hybridization solution (900 mM of NaCl, 20 mM of Tris, pH 7.5, 0.01% of SDS, 20% (vol/vol) of formamide, each probe at a final concentration of 0.1 µM) was dropped on biofilm samples and stained at 46 °C for 4 hours in a chamber humidified with 20% (vol/vol) formamide. Samples were then gently washed in wash buffer (215 mM of NaCl, 20 mM Hydrogen Peroxide Assay. H 2 O 2 concentration was measured by Hydrogen Peroxide Assay Kit (Red Hydrogen Peroxide/Peroxidase Assay Kit, Amplex ™ ). Operations followed a standard protocol of the kit. For testing H 2 O 2 concentration in biofilm supernatant, 80 ul of biofilm supernatant was centrifuged. Subsequently, 50 ul of supernatant was mixed with Hydrogen Peroxide Assay solution. After 30 minutes of reaction, the fluorescent signal (excitation 560 nm/emission 590 nm) was recorded by a Synergy H1 Hybrid Reader. The preparation of a standard curve for quantifying H 2 O 2 concentration followed the standard protocol of the kit. To get data presented in Fig. 4B, Pg, A. actinomycetemcomitans strains and Ss were grown for 48 hours, 24 hours and overnight respectively in TSB to early stationary phase under anaerobic conditions. Cells were resuspended in fresh CDM and 10% inoculated into fresh CDM to get bacterial suspensions. 50 ul of bacteria suspensions were mixed with 50 ul of Hydrogen Peroxide Assay solution and incubated under laboratory atmospheric conditions at 37 °C using the Synergy H1 Hybrid Reader. The optical density (OD 600 ) for cell growth and fluorescent signal for H 2 O 2 concentration were monitored continuously by the reader. For testing H 2 O 2 concentration in the transwell system (96 Well Permeable Support System transwell, Corning ™ HTS Transwell ™ ), Aa and Ss cells were grown to early stationary phase in TSB medium, followed by resuspension in fresh CDM. 50 ul of bacteria suspension and 50 ul of Hydrogen Peroxide Assay solution were mixed at the bottom of the well. The insert was filled with 50 ul of CDM or 50 ul of Aa suspension. After 30 minutes of reaction, the insert was discarded and the H 2 O 2 concentration at the bottom of the well was measured. Three replicates were performed to calculate the means and standard deviations. statistical analysis. All data were obtained from at least three biological replicates. Student's t-test was applied to analyze data on biofilm biomass, H 2 O 2 concentration and CFU.