key: cord-0429957-faas5yq5
authors: Gogos, Artemis; Federle, Michael J.
title: Colonization of the murine oropharynx by Streptococcus pyogenes is governed by the Rgg2/3 quorum sensing system
date: 2020-06-30
journal: bioRxiv
DOI: 10.1101/2020.06.30.179903
sha: 52c7cbd95c3c7a6c2a92d36c89f4de33986684cd
doc_id: 429957
cord_uid: faas5yq5

Streptococcus pyogenes is a human-restricted pathogen most often found in the human nasopharynx. Multiple bacterial factors are known to contribute to persistent colonization of this niche, and many are important in mucosal immunity and vaccine development. In this work, mice were infected intranasally with transcriptional regulator mutants of the Rgg2/3 quorum sensing (QS) system—a peptide-based signaling system conserved in sequenced isolates of S. pyogenes. Deletion of the QS system’s transcriptional activator (Δrgg2) dramatically diminished the percentage of colonized mice while deletion of the transcriptional repressor (Δrgg3) increased the percentage of colonized mice compared to wild type. Stimulation of the QS system using synthetic pheromones prior to inoculation did not significantly increase the percentage of animals colonized, indicating that QS-dependent colonization is responsive to the intrinsic conditions within the host upper respiratory tract. Bacterial RNA extracted directly from oropharyngeal swabs and evaluated by quantitative RT-PCR subsequently confirmed QS upregulation within one hour of inoculation. In the nasal-associated lymphoid tissue (NALT), a muted inflammatory response to the Δrgg2 bacteria suggests that their rapid elimination failed to elicit the previously characterized response to intranasal inoculation of GAS. This work identifies a new transcriptional regulatory system governing the ability of S. pyogenes to colonize the nasopharynx and provides knowledge that could help lead to decolonization therapeutics. Author Summary Streptococcus pyogenes is responsible for a wide spectrum of diseases ranging from common pharyngitis to infrequent invasive infections like necrotizing fasciitis. The ability of this microorganism to persist in the human oropharynx predisposes colonized individuals to a variety of superficial and invasive diseases which lead to significant morbidities and mortality. Identification of the regulatory systems that augment the bacteria’s ability to colonize the oropharynx provides potential targets against which molecular therapeutics can be designed. Here we show that the Rgg2/3 quorum sensing system, an interbacterial communication system, governs the ability of S. pyogenes to colonize the murine oropharynx. Disruption of the system’s transcriptional activator reduced colonization dramatically, eliminated the transcription of two sets of genes known to be activated by the Rgg2/3 system, and tempered the innate immune response seen when S. pyogenes is intranasally infected into the mouse.

Streptococcus pyogenes, or Group A Strep (GAS), is a human-restricted pathogen able to colonize and adapt to numerous tissues of the body. The oronasopharynx is regarded as the most common site of colonization by GAS, which can lead to acute infections such as streptococcal pharyngitis. GAS is responsible for 600 million two operons that include the genes encoding the SHPs, and thus, the net effect of SHP binding to Rgg2 and pheromone production. An additional regulatory element of this circuit is the PepO peptidase, a CovRS-regulated 75 gene that functions to digest the peptide pheromones and turn off the transcription of Rgg2/3 targets [21] .

Initial activation of the Rgg2/3 system can be stimulated by certain environmental conditions, such as low 77 levels of iron and manganese, or the presence of mannose as the primary carbon source [22] . There is evidence 78 for each of these environmental conditions to be present in the pharyngeal mucosa. Metals are generally 79 considered to be restricted by the host and sparingly available to bacteria [23] , although to our knowledge, exact 80 mucosal measurements of iron and manganese have not been studied. However, it is known that mannose is 81 present in the Man3GlcNAc2 core of N-glycan decorations on the surface of nasopharyngeal epithelial cells [24] .

S. pneumoniae is able to break down these decorations using a number of mannosidases [25] , of which 83 homologs exist in GAS [26] . Additionally, it has been confirmed that GAS pharyngeal isolates from cynomologous 84 macaques upregulate mannose catabolism genes, indicating that mannose is likely available to GAS in the 85 pharynx [27] . Due to these environmental conditions, we have previously hypothesized that the Rgg2/3 86 regulatory system may be playing an important role in host-bacterial interactions [21, 22, 28] . In this work, we 87 present evidence showing that the Rgg2/3 QS system governs pharyngeal colonization in the mouse via 88 activation of its downstream gene targets.

PepO degrades SHP pheromones, we reasoned that SHP-dependent signaling activity would remain at low 98 levels until the system can be activated by a favorable environment [21] .

To confirm QS activity in MGAS315, RNA levels of the stcA gene-one of the main targets of the Rgg2/3 00 system-were used as a reporter of QS activity and measured by quantitative RT-PCR (qRT-PCR) from wild 01 type and isogenic strains containing deletions of either rgg2 or rgg3. Cultures were grown in a chemically defined 02 medium (CDM) to determine the baseline activity of the quorum sensing system prior to inoculation of the mouse.

Without the addition of synthetic pheromone, all three strains displayed similar, low levels of expression of the 04 stcA gene (Fig 2A) . However, addition of synthetic pheromone (SHP-C8) to cultures led to rapid and robust 05 induction of the stcA gene in wild type and Δrgg3, consistent with the model that Rgg3 is not required for 06 activation. Conversely, the Δrgg2 mutant was not able to turn on QS targets, regardless of the addition of 07 pheromone, again in line with the idea that Rgg2 is required for activation.

We then examined the importance of an intact Rgg2/3 QS system during murine oropharyngeal 09 colonization by inoculating CD1 mice intranasally with uninduced wild type, Δrgg2, or Δrgg3 bacteria and were among the highest and in the largest proportion of mice for the Δrgg3 strain, it stands to reason the 18 probability for invasive infection were highest in these mice.

Although the Rgg2/3 system is conserved in all sequenced strains of GAS, signaling dynamics and efficiency is not equivalent among strains tested [19, 21, 22] .Therefore, two other GAS serotypes were tested to M49) and HSC5 (M14), and their isogenic rgg2 and rgg3 mutants, were inoculated in mice and monitored 23 periodically by throat swab and CFU counting. Although these serotypes were less informative than MGAS315 24 due to differences in burden and length of bacterial infection, results indicated that the Δrgg2 derivatives were 25 attenuated in their ability to colonize and confirmed that the system governs the same responses in other clinical 26 isolates of S. pyogenes (Fig S1) . Furthermore, these additional strains confirmed that the Δrgg3 mutants 27 colonized better than WT.

Pre-activation of the QS system before inoculation does not affect colonization outcome.

We next tested whether wild type bacteria would colonize to levels similar to the Δrgg3 mutant if they 31 were pre-stimulated with pheromone prior to inoculation. We reasoned that pre-stimulation might enhance the 32 conditioning of bacteria for pharyngeal colonization, producing proteins that would alter the bacterial surface or 33 physiology. We repeated the colonization experiments as described above, with the addition of a cohort of mice 34 inoculated with QS-activated cultures that had been incubated with SHP-C8 pheromone for one hour prior to 35 inoculation. When stimulating MGAS315 cultures for this amount of time, the QS target genes are induced by 36 nearly 1000-fold ( Fig. 2A) . Following nasal inoculation, throat swab monitoring over a nine-day period showed 37 no significant change in colonization between pre-stimulated and unstimulated cultures (Fig. 3) , indicating that 38 the status of QS activity prior to inoculation did not affect the establishment of colonization or mortality.

Intranasal bacterial inoculation leads to the induction of the Rgg 2/3 QS system.

Given the above evidence that in vivo conditions impact quorum sensing, we were eager to measure the 42 activity of the QS system of bacteria within the host. As an initial means to visualize QS activity in living animals,

we inoculated mice intranasally with unstimulated MGAS315 wild type, Δrgg2, or Δrgg3 strains containing a 44 multicopy Pshp3-luxAB reporter before undergoing anesthesia for immobilization. We then obtained imaging at 45 15-and 60-minutes post-infection using an In Vivo Imaging System (IVIS). Imaging showed that luciferase inoculated with the Δrgg2 mutant displayed a lower luciferase intensity and the amount did not change over time (Fig. 4B) .

To validate imaging results and to obtain a more precise quantification of transcription of QS-regulated 50 genes during early stages of colonization, we isolated bacterial RNA directly from mouse throat swabs and 51 conducted qRT-PCR. Three groups of ten mice were inoculated with either wild type, Δrgg2, or Δrgg3. One hour 52 after inoculation, swabs of each group were combined and extracted with TRIzol to isolate nucleic acid. To 53 evaluate QS activation, we assessed transcript levels (normalized to the housekeeping gene gyrA) of two QS-54 targeted genes, stcA and aroE.2, from swab samples and compared them to transcript levels from the uninfected 55 inoculum (Fig. 4D, 4E) . As expected, both stcA and aroE.2 were highly upregulated in wild type and Δrgg3 56 strains that were induced with pheromone in vitro. In bacteria collected from mice, we observed a similar pattern, 

As the IVIS and qRT-PCR data suggested that the Rgg2/3 system is important within the first hour of 63 colonization, we examined the rate of bacterial clearance during the first 24 hours post-inoculation. We 64 performed throat swabs at 0, 4, 8, and 24 hours post-inoculation and showed that even by four hours most bacteria, especially the Δrgg2 cells, were already being eliminated from the oropharynx (Fig. 4F) . The rapid 66 clearance suggests that elimination of bacteria is by a mechanism consistent with humoral components such as 67 antimicrobial peptides (AMPs) and lysozyme, or primary immune mechanisms such mucociliary clearance and 68 swallowing. Although WT and Δrgg3 bacteria were better adapted at surviving this reduction, they too were 69 reduced by several orders of magnitude over 24 hours, congruent with data from 

Quorum sensing non-compliance in the bacterial population does not alter colonization patterns.

Quorum sensing systems fundamentally exist to provide bacteria a mechanism to coordinate gene to establishing colonization, hence a coordinated response by bacteria against host immunity would benefit the population, possibly through a mechanism of coordinated immune avoidance or coordinated assault on immune activities would be a detriment to the population, either because immune responses would become activated or 79 because non-participating bacteria would not contribute to impairing immune activities, especially if they are a colonize. To test these scenarios, mice were co-inoculated with both Δrgg2 and Δrgg3 mutants at a 1:1 ratio and 87 pharyngeal swabs were taken over a nine-day period. Surprisingly, neither predicted outcome was observed, 88 and instead, bacterial colonization by each mutant mirrored the results of when strains were inoculated 89 individually. By the first day, mice were nearly exclusively colonized by Δrgg3 bacteria, whose numbers 90 continued to expand over the first week; Δrgg2 bacteria were cleared by day 3 (Fig. 5A) . Co-inoculations were 91 also carried out for mixtures of wild type with Δrgg2, which also led to the elimination of Δrgg2 bacteria. This 92 combination saw wild type bacteria colonize a moderately higher percentage of the mice, and to higher CFU 93 levels, than the single infection experiment, although these results were not statistically significant (Fig. 5B) . Co-94 inoculation of wild type with Δrgg3 bacteria also showed no statistically significant differences in colonization 95 patterns from those observed in the single infections (Fig. 5C) . Thus, results from these experiments point to a 96 more complex situation occurring in the host environment. One possible confounding factor is that Δrgg2 bacteria 97 are eliminated more rapidly than wild type or Δrgg3, and this rapid removal may not provoke an immune response 98 to an extent where there is a negative impact on colonization by wild type or Δrgg3 bacteria.

The rgg2 mutant fails to induce a pro-inflammatory response in the NALT.

To further investigate the impact that Rgg2/3 QS has on host responses in this colonization model, we 02 assessed bacterial loads and cytokine profiles from nasal-associated lymphoid tissue (NALT) of infected animals.

For each of the three bacterial strains, we inoculated ten mice intranasally. At both 24 and 72 hours-postinfection, we swabbed the throats and surgically removed the NALT organ from each mouse. The NALT is a bi-05 lobed organ that was split into two equal halves; one half was homogenized and plated to obtain bacterial counts, 06 and the other was processed to isolate secreted cytokines and chemokines, which were subjected to a bead-07 based multiplex assay for quantification.

Comparison of the CFU counts between the throat swabs and the NALT showed that all three bacterial 09 strains were taken up into this lymphatic organ that drains the oropharynx and nasopharynx. At 24 hours post-10 infection, all three strains were present in the NALT of at least half of inoculated mice, although the Δrgg3 strain 11 colonized in every instance and had on average 10-fold more CFU counts than either wild type or Δrgg2 (Fig.   12   6B ). In the throat, bacterial counts of all strains were low at 24 hours ( Fig. 6A) , as seen in previous experiments.

By day three, the Δrgg2 bacteria were no longer present in the NALT, and the NALT CFU counts mirrored the 14 throat swab counts for all three strains (Fig. 6A, 6B) . Cytokine response to the Δrgg2 cells, when compared to 15 wild type, showed decreased expression of pro-inflammatory cytokines such as ΙL-1β and TNF-α (Fig. 6D ) which 16 have been previously described to be upregulated during GAS nasal colonization [32] [33] [34] . Likewise, the IFN-γ-17 inducible chemokine CXCL9, previously described to increase after GAS inoculation, was also significantly lower 18 in the Δrgg2 inoculated mice when compared to mice with wild type GAS [33, 34]. Generally, it appeared that

Δrgg2 bacteria failed to elicit cytokine responses and, taken together with the observed rapid clearance from 20 pharyngeal and NALT tissues, may indicate that humoral effectors provide sufficient means to handle these 21 bacteria without eliciting an inflammatory response.

Immune responses to Δrgg3 bacteria were more difficult to interpret, as variability among mice was 23 broader; this was primarily due to one mouse displaying unusually high cytokine and chemokine counts, even 24 though all mice were colonized with high levels of bacteria. However, with exception of this mouse, cytokine and 25 chemokine responses to Δrgg3 were generally lower compared to wild type, especially for a number of regulatory 26 cytokines such as IL-3, IL-4, IL-7, and IL-17, which may help to explain the increased number of Δrgg3 cells that 27 are able to survive both in the NALT and the throat.

This work provides evidence that the Rgg2/3 quorum sensing system is required for oropharyngeal 32 colonization by S. pyogenes. Congruent with our findings, a previous report showed that an insertional disruption 33 in rgg2 decreased the survival of mice in a model of intraperitoneal infection, indicating that this system may be 34 broadly important for the host-pathogen interaction [35] . Recent evidence has shown that the RopB-SIP quorum sensing system and its cysteine protease product SpeB also contributes to GAS pharyngeal colonization in the 

In order to test the hypothesis that the Δrgg2 mutants were being killed before an inflammatory immune response was provoked, we examined bacterial counts and cytokine/chemokine quantification from the NALT 48 ( Fig. 6) . Previous studies have shown that Il-1β and CXCL9 are upregulated in the NALT within twenty-four 49 hours of intranasal infection with GAS [32, 33]. In a model of skin infection, colonization of wild type and MyD88-50 /-mice showed that the secretion of IL-1β and CXCL9 were dependent on MyD88 [34]. We found that IL-1β, 51 CXCL9, as well as the proinflammatory cytokine TNF-α, were all downregulated in Δrgg2-infected mice but not 52 significantly different between mice infected with wild type or Δrgg3 (Fig. 6D) . IL-1β has been shown to promote 53 nasopharyngeal infection by GAS in the mouse model, and LaRock et al. suggest that this initial immune 54 response and recruitment of neutrophils may be crucial for overcoming microbial interference [37] . The lack of 55 activation of IL-1β in the Δrgg2 bacteria lends credence to the hypothesis that QS-off bacteria are eliminated 56 before eliciting MyD88-dependent activation of neutrophil migration to the NALT. As the Δrgg3 bacteria were cytokines and chemokines in the NALT to higher levels than seen for wild type. Curiously, the Δrgg3 bacteria elicited lower expression levels of several regulatory cytokines (IL-3, IL-4, IL-7) than wild type ( Fig. 6D ) even 60 though mice were colonized with substantially higher CFU counts. Two of the Δrgg3-infected mice had at least suggesting that the Δrgg3 bacteria were affecting immune responses differently from wild type.

The inoculation of mice with mixed cultures of Δrgg2 and Δrgg3 bacteria intended to examine the 64 consequences to GAS colonization when a portion of population (Δrgg2) is unable to respond to pheromones produced by co-inoculated QS-active bacteria. Perplexingly, results of the mixed-culture inoculation experiments 66 also showed that Δrgg3 mutants, which colonized to high levels in the pharynx, provided no benefit to Δrgg2 67 colonization (Fig 5A) . These findings lead us to question how QS, which conceptually evolves to exert behaviors 68 of cooperativity among community members, was in this case only benefiting GAS participating in QS. The 69 exclusion of Δrgg2 mutants is reminiscent of QS policing mechanisms seen in Pseudomonas aeruginosa, where 70 non-cooperative bacteria, or cheats, that exploit benefits shared by the community would incur a threat to the 71 population if left unchecked [39, 40] . QS regulons that express "private good" genes, such as niche-specific 72 catabolic pathways or stress response systems, confer advantages to individual bacteria, but only those that 73 comply with social activities; bacteria that do not respond to pheromones or autoinducers (cheaters) will not 74 induce these private goods and will therefore be less fit. Identifying the underlying mechanism via which Δrgg2 75 bacteria are unable to establish colonization remains to be determined. Due to this finding, the identification of is not a case of mere genetic redundancy but rather is important to the regulatory system's architecture. The 83 data in this work confirms the importance of both regulators, showing that the deletion of either rgg2 or rgg3 84 causes in vivo phenotypic divergence from wild type GAS behavior. Wild type bacteria appear to finely tune the expression of this system by using both regulators and the products produced by the system make great potential 86 targets for vaccine development. Table 1 . Streptococcus pyogenes NZ131, MGAS315, and HSC5 were routinely cultivated in a chemically defined 92 medium, CDM, as described previously [19] at 37 o C and 5% CO2, without shaking. In order to facilitate timing 93 of the experiments, starter cultures were prepared. Strains were grown overnight in THY broth, and in the 94 morning, cultures were diluted 1:100 into CDM and grown to mid-logarithmic phase (OD600=0.5 to 0.6). Glycerol 95 was added to a final concentration of 20% and aliquots of this culture were frozen and stored at -80° C. On 96 experiment days, aliquots were diluted into fresh CDM 1:20 and grown to densities as specified below. In the 97 study where bacteria were pretreated with synthetic peptide to activate the QS system before inoculation into 98 the mouse, 100 nM SHP-C8 pheromone was added to the culture one hour before collection for inoculation.

Construction of bacterial strains.

The HSC5 Δrgg2 and Δrgg3 strains were constructed using the methods previously utilized to make the 

6AB of the main text. Gogos and Federle, "Colonization of the murine oropharynx by Streptococcus pyogenes is governed by the Rgg2/3 quorum sensing system" Gogos and Federle, "Colonization of the murine oropharynx by Streptococcus pyogenes is governed by the Rgg2/3 quorum sensing system" 

Deceased mice were not included during the calculation of percent colonized mice. A Friedman test was again 30 run to compare the colonization percentages between the three strains, and the colonization rates of each Gogos and Federle, "Colonization of the murine oropharynx by Streptococcus pyogenes is governed by the Rgg2/3 quorum sensing system" pheromone-treated groups. The colony counts on each day were compared using multiple t-tests (α≤0.05).

Gogos and Federle, "Colonization of the murine oropharynx by Streptococcus pyogenes is governed by the Rgg2/3 quorum sensing system" were compared to the WT counts using a Wilcoxon test (α≤0.05). The Δrgg2 colony counts were statistically Gogos and Federle, "Colonization of the murine oropharynx by Streptococcus pyogenes is governed by the Rgg2/3 quorum sensing system" Gogos and Federle, "Colonization of the murine oropharynx by Streptococcus pyogenes is governed by the Rgg2/3 quorum sensing system" 

Page Figure S1 : Colonization of the murine oropharynx by S. pyogenes strains NZ131 and HSC5 s-2 s-2 SUPPLEMENTAL FIGURE 1. Colonization of the murine oropharynx by S. pyogenes strains NZ131 and HSC5. Figure S1 . Colonization of the murine oropharynx by S. pyogenes strains NZ131 and HSC5. GAS strains NZ131 and HSC5 were analyzed for QS activity in culture and for colonization levels in the mouse. (A/D) Relative transcript levels of QS-target gene stcA in cultures of WT, Δrgg2, and Δrgg3 strains grown in CDM broth without (uninduced) or with (induced) the addition of 100 nM SHP-C8 pheromone. Bars are the mean fold differences compared to uninduced WT for three biological replicates. Asterisks indicate significance of multiple t-tests comparing each sample to uninduced WT, α≤0.05. (B/E) Three-week old CD1 mice were inoculated intranasally with 10 9 CFU of uninduced bacteria. CFU counts obtained from throat swabs of each mouse are plotted for each day of sampling. Skull and crossbones denote the number of animals that displayed humane endpoint signatures requiring euthanasia (signs of distress indicating invasive infection) and are shown in an accumulating manner. Ten mice were utilized for each bacterial strain; bars indicate the median CFU/swab value and the detection limit was 60 CFU/swab. Using the Friedman test, a nonparametric matched ANOVA, the NZ131 strains were determined to have statistically different median CFU counts over the first three days of the experiment, p=0.028. The HSC5 strains were compared for the first day using the Friedman test and had statistically significant median CFU counts, p=0.0015. (C/F) The percentage of mice colonized during the experiment shown in panels B and E were compared between the three strains over time. Deceased mice were not included during the calculation of percent colonized mice. A Friedman test was again run to compare the colonization percentages between NZ131 three strains on the first three days, and the colonization rates of each strain were found to be statistically significant from one another, p=0.0069. The first day's percentage of colonized HSC5 strains were also compared using a Friedman test, no significance was found (α≤0.05).

s-3

Data correspond with Figure 6AB of the main text.

The global burden of group A streptococcal diseases

Roles of Pseudomonas aeruginosa las and rhl quorum-14

Acceleration of Enterococcus faecalis biofilm formation by aggregation 12 substance expression in an ex vivo model of cardiac valve colonization

Quorum-sensing regulators control virulence gene expression in Vibrio cholerae

Quorum sensing and policing of Pseudomonas aeruginosa social cheaters

Genetic and Structural Analyses of RRNPP Intercellular Peptide Signaling of Gram

Genome sequence of a nephritogenic and highly transformable M49 strain of Streptococcus pyogenes

Genome sequence of a serotype M3 strain of group A Streptococcus: phage-encoded toxins, the high-virulence phenotype, and clone emergence

Two group A streptococcal peptide pheromones act through opposing Rgg regulators to control biofilm development

Genome-wide identification of genes required for fitness of group A Streptococcus in human blood

Induction of a quorum sensing pathway by environmental signals enhances group A streptococcal resistance to lysozyme

Work herein was supported by funds associated with NIH R01-AI091779. AG is supported by F30AI136359 and MF