Different Group A Streptococcus pili lead to varying proinflammatory cytokine responses and virulence

The human pathogen Streptococcus pyogenes, or Group A Streptococcus (GAS), is associated with a variety of diseases ranging from mild skin and soft tissue infections to invasive diseases and immune sequelae such as rheumatic heart disease. We have recently reported that one of the virulence factors of this pathogen, the pilus, has inflammatory properties and strongly stimulates the innate immune system. Here we used a range of nonpathogenic Lactococcus lactis gain‐of‐function mutants, each expressing one of the major pilus types of GAS, to compare the immune responses generated by various types of fully assembled pili. In vitro assays indicated variability in the inflammatory response induced by different pili, with the fibronectin‐binding, collagen‐binding, T antigen (FCT)‐1‐type pilus from GAS serotype M6/T6 inducing significantly stronger cytokine secretion than other pili. Furthermore, we established that the same trend of pili‐mediated immune response could be modeled in Galleria mellonella larvae, which possess a similar innate immune system to vertebrates. Counterintuitively, across the panel of pili types examined in this study, we observed a negative correlation between the intensity of the immune response demonstrated in our experiments and the disease severity observed clinically in the GAS strains associated with each pilus type. This observation suggests that pili‐mediated inflammation is more likely to promote bacterial clearance instead of causing disruptive damages that intensify pathogenesis. This also indicates that pili may not be the main contributor to the inflammatory symptoms seen in GAS diseases. Rather, the immune‐potentiating properties of the pilus components could potentially be exploited as a vaccine adjuvant.


INTRODUCTION
Streptococcus pyogenes (Group A Streptococcus, GAS) is an important human pathogen that causes a wide range of diseases including skin and throat infections (e.g.2][3][4] While there are 600 million cases of pharyngitis annually worldwide, the greatest burden is attributed to the immune complications following superficial infection.These conditions include acute rheumatic fever (ARF) and rheumatic heart disease (RHD), which can cause lifelong morbidity and premature deaths.ARF primarily affects children in low-to-middle-income countries and commonly progresses into RHD as these individuals reach their 20s, with estimations of global RHD prevalence at 40.5 million and mortality at 306,000 per year. 5Despite being effective in treating infections, antibiotics have failed to reduce disease burden in less privileged areas where invasive GAS infections still cause high mortality rates, and superficial manifestations continue to consume substantial health care resources. 6pon contact with pathogens, innate immunity launches the fast-acting initial response by sensing pathogenassociated molecular patterns via pattern recognition receptors.Immune cells sensing the microbes produce cytokines, which in turn stimulate further cytokine release in a magnifying manner to activate the infection-specific adaptive responses.Many inflammatory symptoms in infectious diseases are associated with the cytokines produced as a part of the innate immune response to the pathogen.Bacterial infections typically induce adaptive immune responses via interleukin (IL)-12 and interferon c signaling (which facilitates T helper type 1 differentiation) or IL-23 signaling (which promotes T helper type 17 differentiation).However, when these T-cell subsets become predominant, they can contribute to the development of autoimmune disease, causing prolonged inflammation and tissue damage. 7,8In the context of GAS disease, the cytokine and chemokine profile of pharyngitis has been revealed in a recent human challenge model, with the elevation of serum IL-1Ra, IL-6, interferon-c, C-X-C motif chemokine ligand 10 (CXCL-10) and IL-18 being an early characteristic of experimental human pharyngitis. 9Furthermore, cytokine and chemokines are part of the key components in autoimmune disease 8 and have been implicated in the pathogenesis of ARF and RHD, as summarized in a recent review by Middleton et al. 10 For instance, tumor necrosis factor (TNF), which has been found to be substantially elevated in the sera of patients with RHD, stimulates further cytokine secretion, amplifying the state of inflammation and sustaining high levels of chemotaxis. 11,12Peripheral TNF has also been linked to other GAS-mediated conditions, such as the disruption of the blood-brain barrier following GAS sepsis. 13A recent review by Wilde et al. 14 exemplifies a number of GAS virulence factors that mediate inflammatory mechanisms in disease manifestations, and illustrates the seemingly paradoxical contributions of these proinflammatory virulence factor mechanisms.However, a fuller picture of the actual bacterial components that contribute to these immune complications remains to be drawn.
We have previously demonstrated that the GAS pilus is a strong immune stimulant, with the ability to induce high levels of proinflammatory cytokines such as TNF and IL-8 via the Toll-like receptor 2 pathway. 15The pilus (plural, pili) is a long, thin, hairlike surface structure associated with increased pathogenesis in mouse models. 16The pilus shaft is mainly formed by the covalently linked backbone pilins.On top of the pilus fiber is the ancillary pilin 1 (AP1), also known as the collagen-binding protein of Group A streptococci (Cpa), which has adhesive properties and serves as the pilus adhesin.All pilus types except fibronectin-binding, collagen-binding, T antigen (FCT)-1 have an additional ancillary protein (AP2), which sits at the base and serves as an anchor to the bacterial cell wall. 17hese pilus structural proteins, along with the pilus sortases, are encoded in an operon located within the variable FCT region.Since the discovery of GAS pili in 2005, nine pilus types have been described with varying structures and characteristics, 18,19 with six types being distinctly different in terms of antigenicity (Figure 1a).For example, the pili expressed by the serotype M6/T6 strain (PilM6) and serotype M4/T4 strain (PilM4) belong to the FCT-1 and FCT-5 type, respectively, of which the encoding genes are highly variable with $ 75% heterogeneity. 191][22][23] Interestingly, GAS strains associated with distinct clinical presentations often possess different pilus variants. 4,24While the serotype M1/T1 strains are associated with more severe diseases and epidemics, 25,26 the M6/T6 strains are often seen in self-limiting conditions such as pharyngitis. 24e hypothesize that the immunostimulatory properties of the pilus may contribute to the sustained inflammatory environment and disease progression.Because of the highly diverse nature of pili, it is possible that different pilus types possess varying abilities to promote inflammation, which in turn exacerbates disease progression.In this study we focused on three different pilus types produced from an M28/T28.1 strain associated with ARF, 27 an M1/T1 strain linked to invasive diseases 26 and an M6/T6 strain associated with mild pharyngitis. 28We also included pili from M2/T2 and M4/T4 strains to cover the five main FCT types (FCT-3/4, FCT-2, FCT-1, FCT-6 and FCT-5, respectively) grouped by genetic clustering (Figure 1b).We used a collection of gain-of-function Lactococcus lactis strains expressing different GAS pilus operons to assess the proinflammatory abilities of each pilus type in cultured immune cells and a Galleria mellonella infection model.Galleria mellonella (the greater wax moth) has become an increasingly popular model to study microbial infections, as it can be easily manipulated and maintained, is cost effective and has greater ethical acceptance compared with mammalian animal models. 28][30] The model has an innate immune system that produces a remarkably similar responses to vertebrates, which include six types of phagocytic cells, namely, prohemocytes, coagulocytes, spherulocytes, oenocytoids, plasmatocytes and granulocytes. 31These hemocytes are responsible for mediating the immune response of G. mellonella, with functions aside from phagocytosis such as clotting of hemolymph, encapsulation and forming aggregates to trap pathogens. 32As reported previously, circulating hemocyte counts increase following infection and increasing hemocyte density reflects the pathogenicity of the infecting pathogen. 33,34This is reminiscent of the human response to pathogens.Thus, shifts in G. mellonella hemocyte number were used to evaluate the response to different GAS pili on L. lactis.

GAS pili induce TNF secretion from monocytes
To investigate the immune responses against different types of GAS pili, we utilized L. lactis, a nonpathogenic Gram-positive species related to GAS, as a vehicle to present fully assembled pili to cultured immune cells.In brief, each pilus operon was cloned and expressed in L. lactis under the control of the constitutive lactococcal P23 promoter as previously described. 35Flow cytometry analysis confirmed consistent pilus expression between  strains. 36Wild-type (WT) Lactococcus lactis and five L. lactis gain-of-function strains with cell surface expression of different GAS pilus types were incubated with human monocytic THP-1 cells for 20 h, and the secreted proinflammatory TNF cytokine was quantified by ELISA.Each of the five investigated pilus types increased TNF release from THP-1 monocytes (Figure 2).Compared with L. lactis WT, the increase was 3.2-fold for L. lactis PilM1 (P = 0.03), 3.8-fold for L. lactis PilM2 (P = 0.02), 3.5-fold for L. lactis PilM4 (P = 0.02), 6.3-fold for L. lactis PilM6 (P < 0.001) and 3.4-fold for L. lactis PilM28 (P = 0.03).

GAS pili demonstrated varying virulence in a G. mellonella infection model
Contrary to our expectation, the pilus type showing the strongest cytokine-inducing ability in the THP-1 cell experiment was PilM6, which is expressed by the M6/T6 strain that often presented in mild, localized infections.By contrast, pilus from the M1/T1 GAS strain, which is highly associated with severe invasive infections, only induced moderate level of TNF among the panel of tested pilus types.To verify this observation in an in vivo infection model, we compared the virulence of GAS pili using the same panel of gain-of-function L. lactis strains in a G. mellonella larvae infection model.The similarities between the innate immune system of these wax moth larvae and vertebrates make them a useful in vivo model for inspecting inflammatory responses to microbial complexes such as the pilus.
A dose titration was first performed for the whole panel of pili-expressing L. lactis to determine the optimal inoculum dose.G. mellonella larvae were infected with five different doses, and PBS-injected larvae were used as a control.During the 5-day monitoring period, all PBSinjected control larvae survived the entire time course.By contrast, all strains of pilus-expressing L. lactis caused lethality at higher dosages, with 1 9 10 9 CFU resulting in 100% larval death by the end of the monitoring period.Larvae administered 2 9 10 8 CFU also experienced high mortality, with a 90-80% death rate observed on day 5 (Figure 3a).Inoculation of WT L. lactis resulted in a slightly dampened but not significant lethality in larvae compared with its pilus-expressing counterparts, with survival of 10% and 30% of larvae inoculated with 1 9 10 9 CFU and 2 9 10 8 CFU, respectively (Table 1).When the dosage was lowered to 4 9 10 7 CFU, the survival rate of larvae given the WT bacterium rose to 80%, whereas larvae administered with pili-expressing L. lactis had a slightly lower survival rate of 70-60% (Table 1).The 5-day survival curves of each L. lactis strain across different inoculating CFUs are provided in supporting information (Supplementary figure 1).
As indicated by the nonlinear regression curves for larval death (Figure 4), the log median lethal dose of WT   it appeared that dosages in the 1 9 10 7 CFU range would be optimal to stimulate a response without overwhelming the immune system and leading to widespread mortality.
A health index scoring system that encompasses attributes associated with immune response such as melanization has been established for G. mellonella (Table 2).This scheme was used to measure the pathogenicity of different L. lactis strains in larvae that remained viable following inoculation with different doses of the bacteria.In general, health index scores matched the survival curves, with increasing L. lactis dosage correlating with decreased health score (Figure 3b).In addition, health index scores provided a mechanism to detect more refined differences in larvae health.For example, the survival rate of larvae administered PilM2, PilM4 and PilM28 at the 4 9 10 7 CFU dosage was all 70%.However, when looking at the health index scores at this dose, it became apparent that PilM28 induced a greater health burden on the larvae, with the day 5 score trailing behind PilM2 and PilM4 by 0.55 and 0.75 points, respectively (Table 1).In addition, on day 5 where all larvae survived inoculation with L. lactis strains, the health index score of larvae administered WT bacterium was up to 1.35 points higher than larvae exposed to pilus-expressing L. lactis (Table 1).Individual health index score data for each L. lactis strain over the 5-day monitoring period are provided in Supplementary figure 2.

Hemocyte density of G. mellonella reflect the proinflammatory activity of GAS pili
Health index scoring is a useful tool to capture differences in immune response to pili-expressing L. lactis, especially at nonlethal doses where subtle differences in survival rates can be observed.However, the health index relies on changes to external appearance, which can take 24 h to develop.To acquire information that can reflect more immediate immune responses following exposure to pili, we recognize the need for a complementary approach.Circulating hemocyte density after infection can provide an early indication of the immune response in G. mellonella.Previous studies have shown marked increases in circulating hemocyte as early as 4 h after infection with bacteria. 34In addition, a hemocyte count provides a more objective measurement to illustrate immune response compared with the health index, which relies on subjective visual inspections.Hemocyte density was therefore compared between G. mellonella larvae injected with WT L. lactis and gain-of-function pilus-expressing strains.A nonlethal infection dose of 1 9 10 7 CFU was chosen based on results shown in Figure 3, and hemocyte density was assessed at multiple time points over 3 days.Hemocyte density of sham-injected larvae was used as a baseline and subtracted from the cell counts.The differences in hemocyte density was most prominent at 6 h after infection, where the pili-expressing bacteria appeared to induce a surge in circulating hemocytes.At this time point, G. mellonella larvae administered with pili-expressing strains had a significantly higher hemocyte density than larvae administered with WT bacterium (Figure 5), with mean cell count being between 1.5-fold (P = 0.01) and 2.5fold (P < 0.001) greater.The hemocyte density of larvae administered with the WT strain did not differ significantly from larvae receiving PBS (P > 0.99).In line with the previous assays studying THP-1 cell TNF response (Figure 2), PilM6 in particular seemed to stimulate especially high hemocyte counts 6 h after infection, with cell counts between 1.4-fold (compared with PilM2, P < 0.001) and 1.6-fold (compared with PilM1, P < 0.001) greater than those larvae administered with the other piliexpressing strains.This trend was preserved at the 24-h time point, with pili-expressing bacteria continuing to induce significantly higher hemocyte density in larvae than those with WT bacterium, and larvae exposed to PilM6 showing significantly greater hemocyte counts than those exposed to other pilus types.While hemocyte density started to drop past this time point, larvae inoculated with PilM6 L. lactis continued to maintain a hemocyte density significantly higher than that in the WT L. lactisinoculated larvae, until the end of the time course (P < 0.001; Figure 5).

Lactococcus lactis expressing different GAS pili were cleared at different rates in G. mellonella
Observations on the differing hemocyte densities following inoculation with different pili-expressing L. lactis posed the question of how these changes in the wax moth immune system affected the clearance of L. lactis during the course of this infection.To elucidate the relationship between hemocyte density and bacterial clearance, G. mellonella larvae were infected according to the hemocyte assay and after 0, 24, 48 and 72 h, larvae were homogenized and plated to perform colony enumeration.The CFUs of bacteria recovered from the larvae groups subjected to differing inoculations began to diverge 24 h after inoculation.In particular, piliexpressing L. lactis were cleared much faster than WT bacteria, with colony counts recovered from larvae given pili-expressing strains between 1.6-fold (PilM2, P = 0.001) and 5.2-fold (PilM6, P < 0.001) lower than those of larvae administered with WT strain.PilM6expressing L. lactis experienced an especially fast clearance, with recovered CFU significantly lower than larvae inoculated with PilM1-, PilM2-or PilM4expressing bacteria (P < 0.01, Figure 6).These results are congruent with the hemocyte density assay, with the larvae inoculated with pili-expressing L. lactis consistently showing higher hemocyte density counts and faster bacterial clearance 24 h after infection, while those exposed to WT bacteria having lower hemocyte counts and slower bacterial clearance.This indicates that pili promote an enhanced immune response, resulting in faster clearance of the bacteria.Larvae administered with PilM6-expressing L. lactis demonstrated the highest surge in hemocyte count after exposure, and inversely the fastest rate of bacterial clearance over this time frame (Figure 6).

DISCUSSION
Since the discovery of GAS pili, most studies of this surface-abundant structure have focused on its role as a virulence factor and potential as a vaccine target (reviewed in Nakata and Kreikemeyer 37 ), while its impact on immune responses has not been well-studied.The antigenic differences between the various GAS pilus types pose questions about the differences in immune response following infections with strains possessing different pilus types.To examine the inflammatory properties of different GAS pili, there was a need to isolate the Figure 5. Group A Streptococcus pili stimulate an increase in hemocyte density in G. mellonella.Wax moth larvae were injected with 1 9 10 7 colony-forming units (CFU) of wild-type (WT) Lactococcus lactis or L. lactis gain-of-function strains expressing PilM1, PilM2, PilM4, PilM6 and PilM28.Hemolymph was collected from larvae at 6, 24, 48 and 72 h following inoculation and counted using a hemocytometer.A hemocyte count was also obtained from a sham-injected control group and subtracted from all cell density counts.Experiments were performed in duplicate and data from three independent biological repeats are shown as mean AE standard deviation.Statistical significance was determined by two-way ANOVA.P-values were calculated by Holm-S ıd ak's multiple comparisons test and are shown in the table.PBS, phosphate-buffered saline.
structure from other GAS virulence factors that could also direct immune responses and affect the inflammatory outcome.Lactococcus lactis provides several advantages as a surrogate in this aspect.It is a nonpathogenic bacterium related to GAS with available genetic manipulations tools, and the two species share similar mechanisms for processing surface proteins. 38Our results show that the heterogeneous expression of GAS pili gave L. lactis an increased ability to stimulate TNF production in the monocytic THP-1 cell line.Using the preexisting framework for inspecting bacterial virulence in the G. mellonella infection model, the gain-of-function L. lactis strains also demonstrated enhanced pathogenic characteristics in this nonvertebrate model, although this nonpathogenic surrogate required a much higher inoculum than GAS to show a notable change in health index. 30The presence of the GAS pilus complex appeared to increase the immune response and health burden of the wax moth larvae, even with the PilM1-expressing strain which has a low stimulatory effect in the THP-1 cell-based cytokine assay.This was demonstrated by inoculations with the piliated strains resulting in increased hemocyte density, and worse health index scoring in the G. mellonella larvae.
This study demonstrated that there were variations in the proinflammatory responses to different pili.However, this did not correspond to the disease severity associated with isolates bearing this particular pilus type.In our panel, PilM6 appears to stimulate a significantly higher inflammatory response than other pilus types, although this FCT-1 pilus was derived from the M6/T6 strain associated with noninvasive, self-limiting pharyngitis infections. 24By contrast, the FCT-2 pilus (PilM1), which showed the lowest immune response in this study, was associated with serious clinical presentations such as invasive infection.Furthermore, our results show that PilM28 did not elicit particularly high inflammatory responses in THP-1 cells or cause intense immune response in the G. mellonella infection model, despite FCT-3/4 pilus types being commonly expressed by GAS strains associated with ARF. 4 Emerging research studies have described the complex and sometimes seemingly paradoxical effects of proinflammatory stimulations from GAS (reviewed in Wilde et al. 14 ).An inflammatory reaction is required for the coordination of antimicrobial processes to resolve bacterial infections; however, this is also one of the driving forces behind tissue damage in the progression of invasive GAS infections. 39In the case of GAS pili, the results of this study indicate that pilimediated inflammation is more likely to initiate an alarm signal for the immune system to act on clearing the pathogen, rather than promoting disruptive mechanisms such as tissue damage that intensify pathogenesis.This assumption is supported by the bacterial enumeration experiments (Figure 5) presented in this study showing increased clearance in strains expressing GAS pilus.While organisms with very high pathogenicity can overwhelm the larval immune response and result in a reduction of hemocytes with ongoing growth of the pathogen, 40 this was not observed in the case of L. lactis expressing GAS pilus.At the dosage used in this study, the presence of pili corresponded to increasing hemocyte density, but the surge of hemocytes following infection could mount an effective immune response to start clearing the bacteria.Comparing bacterial enumerations from larvae injected with different L. lactis strains, WT bacteria generated a less pronounced hemocyte response and in turn had a slower rate of bacterial clearance, while piliexpressing bacteria prompted greater hemocyte response and a corresponding higher rate of bacterial clearance.These results again imply that the inflammation induced by GAS pili are unlikely to contribute to severe clinical presentations.
The conclusion that GAS pili are proinflammatory immune stimulants that are not likely to cause severe disruptive effect also implies that incorporating pili in vaccine development can be an effective way to trigger the immune system without serious safety concerns.6][47] Historical trials using a high dose of partially purified M protein in vaccination have been associated with the development of ARF. 48,49y contrast, the pili are not implicated in the direct pathogenesis of ARF/RHD and have not been implicated in the production of autoantibodies.Homology searches of the pilus proteins against the human proteome have also been conducted and revealed low identity with no hits for proteins of the heart. 4Our study emphasized that the inflammatory properties of pili alone do not contribute to significant pathogenicity but may enhance host immune responses.This suggests that a pilus-based vaccine can induce strong antibody response without causing damage to the recipient.Overall, our investigation showcases that there are variations in the inflammatory and immune responses induced by different GAS pilus types.We also demonstrated the ability of G. mellonella larvae to be utilized as a tool to investigate immune response toward GAS virulence factors expressed on L. lactis.While all the L. lactis strains expressing GAS pili demonstrated immune-stimulating effects in comparison to the nonpiliated WT bacterium, PilM6 elicited a significantly higher inflammatory response in THP-1 cells and distinctively high levels of circulating hemocytes in the G. mellonella infection model.To date, most research on GAS pili focuses on their function in host adherence and colonization, while the immune-related features have only recently been recognized.We recently identified the FCT-2 pilus (PilM1) as a Toll-like receptor 2 ligand with immune-potentiating function. 15When extending the investigation to include more GAS pilus types, we also found that the pilus subunit Cpa from PilM6 stimulated the highest level of Toll-like receptor 2 activation compared with those from PilM1, PilM2, PilM4 and PilM28 (unpublished data).Of note, previous studies investigating GAS pili presented on L. lactis have indicated that pilus expression levels on the surrogate bacteria are uniform across different pilus types, including those examined in this study. 36This suggests that the differences in the immune response to various pili expressing L. lactis strains observed here were not a result of variability in the amount of pili available to interact with immune cells, and were likely a consequence of diversity in the structure and functional properties across pilus types.The structure of FCT-1-type pili such as PilM6 is the most divergent from the other FCT-type pilus complexes. 18Previous investigation into this pilus type revealed its high biofilm formation activity, which unlike other FCT types was not dependent on cell culture conditions. 50,51PilM6 was determined to contribute to this formation of bacterial aggregates via the Cpa. 52It is possible that PilM6-promoted bacterial aggregation increased the contact of PilM6-expressing L. lactis with the co-incubated THP-1 cells, which thus contributed to the higher TNF release recorded from the panel tested in this assay.A similar phenomenon may have occurred when the bacterium was injected into the G. mellonella larvae, accounting for the equally high immune response.Although the molecular details of pili-induced proinflammatory remain to be investigated, and the actual linkage between pilus types and disease severity awaits to be ruled out in the clinical setting, our results indicate that the immune-stimulating property of GAS pilus alone is unlikely to trigger severe inflammation.By contrast, the proinflammatory response to the GAS pilus could be exploited in vaccine development such as a novel adjuvant system.

Bacterial strains and growth conditions
Bacterial strains used are listed in Table 3.All L. lactis MG1363 strains used were created in a previous study, in which consistent pilus expression across the strains was confirmed by flow cytometry. 36Bacteria were grown overnight under static conditions at 28°C in M17 broth (BD Biosciences, Franklin Lakes, NJ, USA) supplemented with 5% glucose (GM17) and 200 lg mL À1 kanamycin where appropriate.For the G. mellonella infection model, cells were resuspended in PBS with 10% glycerol and snap frozen.Before experiments, an aliquot of each strain was thawed and cultured on GM17 agar plates for colony enumeration to determine and prepare the appropriate dosages for injection.

Cytokine assay
Lactococcus lactis strains, either WT or expressing pilus from GAS serotypes (PilM1, PilM2, PilM4, PilM6 and PilM28), were grown overnight in GM17 medium before resuspension in RPMI-1640 medium to a cell density of approximately 1 9 10 7 CFU mL À1 (calculated using the conversion equation optical density measured at a wavelength of 600 nm [OD 600 nm ] of 1.0 = ~1 9 10 8 CFU mL À1 ).THP-1 cells were seeded into 96 well plates at a density of 1 9 10 5 cells per well in RPMI-1640 medium and treated with 1 9 10 5 L. lactis to achieve a multiplicity of infection of 1.The Toll-like receptor 2 agonist Pam 3 CSK 4 (InvivoGen, San Diego, CA, USA) was used as a positive control.The plate was incubated at 37°C in the presence of 5% CO 2 for 20 h prior to the collection of culture supernatant and determination of cytokine levels using Crux Biolab human cytokine ELISA kits (Bayswater, VIC, Australia) as per the manufacturer's instructions.

Galleria mellonella infection model
Antibiotic-free G. mellonella larvae were purchased from Biosuppliers (Auckland, New Zealand), maintained in the dark at room temperature in original packaging with feed and used within 2 weeks after arrival.Larvae (n = 5) of the size between 1.5 and 2 cm, with no visible melanization, were chosen for each experimental group.Five different L. lactis strains expressing pilus from GAS serotypes (PilM1, PilM2, PilM4, PilM6 and PilM28) and WT L. lactis were injected at 1 9 10 9 CFU, 2 9 10 8 CFU, 4 9 10 7 CFU, 8 9 10 6 CFU and 4 9 10 5 CFU in a volume of 20 lL into the lower left proleg of the larvae using an insulin syringe (BD Biosciences, Franklin Lakes, NJ, USA).The control group received 20 lL PBS.After injection, each larva was housed in individual wells of a 12-well plate and incubated at 37°C without food and monitored over 5 days.Every 24 h, larvae health was recorded using a previously established health index score 30 based on activity, cocoon formation, melanization and survival (Table 2).

Hemocyte density
Twenty-four G. mellonella larvae (each ~1.5 cm in length) were selected for each experimental group.Ten million CFU WT L. lactis or gain-of-function strains expressing pilus from GAS serotypes (PilM1, PilM2, PilM4, PilM6 and PilM28) were injected into the lower left proleg of the larvae in a volume of 20 lL using an insulin syringe.The control group received 20 lL PBS and another group of larvae were sham injected to calculate the baseline hemocyte mobilization elicited by injury.Subsequent to injection, each larva was housed in individual wells of a 12-well plate without food and incubated at 37°C.At each postinfection time point (6, 24, 48 and 72 h), hemocytes were collected from six larvae from each group.Larvae were placed on a Petri dish on ice to slow down movement.Once they were no longer motile, each larva was pricked on the lower left proleg with an insulin syringe while applying pressure to the top half of the body, and 10 lL of hemolymph was collected from the puncture using a pipette.Ten microliters of PBS with 0.37% b-mercaptoethanol was added to each hemolymph sample to prevent melanization, and hemocyte density was determined using a hematocytometer.Hemocyte density of larvae that had not undergone inoculation was used as a baseline and subtracted from the cell counts.

Lactococcus lactis enumeration
Twenty-four G. mellonella larvae (~1.5 cm in length) were selected for each experimental group.About 1 9 10 7 CFU of six different L. lactis strains, either WT or expressing pilus from GAS serotypes (PilM1, PilM2, PilM4, PilM6 and PilM28), were injected into the lower left proleg of the larvae in a volume of 20 lL using an insulin syringe.Following injection, each group of larvae was placed in 12-well plates without feed and incubated at 37°C.At 0 (immediately after inoculation), 24, 48 and 72 h after infection, six larvae from each group were mechanically homogenized with a micro tissue homogenizer (Corning, New York, USA) in 0.5 mL PBS and cultured in triplicate on GM17 agar plates overnight at 28°C to conduct colony enumeration.

Figure 1 .
Figure 1.Variation of Group A Streptococcus pilus types.(a) Gene organization of the antigenically different fibronectin-binding, collagenbinding, T antigen (FCT) regions identified thus far.Dark gray arrows are transcriptional regulator genes, blue arrows are fibronectin-binding protein genes, green arrows are previously reported or inferred surface-expressed pilus protein genes, hatched gray arrows are sortase genes or putative genes encoding sortase and light gray arrows are other open reading frames.Adapted from Kratovac et al. 19 (b) Phylogenetic trees based on backbone pilin (BP) (i), collagen-binding protein of Group A streptococci (Cpa) (ii) and ancillary pilin 2 (AP2) (iii) gene sequences.Different variant groups are represented by different colors.Adapted from Falugi et al.18

18
Figure 1.Variation of Group A Streptococcus pilus types.(a) Gene organization of the antigenically different fibronectin-binding, collagenbinding, T antigen (FCT) regions identified thus far.Dark gray arrows are transcriptional regulator genes, blue arrows are fibronectin-binding protein genes, green arrows are previously reported or inferred surface-expressed pilus protein genes, hatched gray arrows are sortase genes or putative genes encoding sortase and light gray arrows are other open reading frames.Adapted from Kratovac et al. 19 (b) Phylogenetic trees based on backbone pilin (BP) (i), collagen-binding protein of Group A streptococci (Cpa) (ii) and ancillary pilin 2 (AP2) (iii) gene sequences.Different variant groups are represented by different colors.Adapted from Falugi et al.18

Figure 2 .
Figure 2. Group A Streptococcus pili induce tumor necrosis factor (TNF) responses in THP-1 cells.THP-1 cells were incubated overnight with wild-type (WT) Lactococcus lactis or L. lactis gain-of-function mutants expressing PilM1, PilM2, PilM4, PilM6 or PilM28 at a multiplicity of infection of 1 before the TNF released into the supernatant was quantified by ELISA.Untreated cells were used as the negative control, and the Toll-like receptor 2 agonist Pam 3 CSK 4 (1 lg mL À1 ) was used as the positive control.Experiments were performed in duplicate and data from three independent biological repeats are shown as mean AE standard deviation.Statistical significance was determined by one-way ANOVA, and the P-values were calculated by Holm-S ıd ak's multiple comparisons test.*P ≤ 0.05, ***P ≤ 0.001 compared with cells incubated with WT L. lactis.

Figure 3 .
Figure 3. Mean death rates and health index scores from 10 G. mellonella larvae per dosage following inoculation with different Lactococcus lactis strains on day 5.Ten G. mellonella larvae per dosage were injected with 1 9 10 9 colony-forming units (CFU), 2 9 10 8 CFU, 4 9 10 7 CFU, 8 9 10 6 CFU and 4 9 10 5 CFU of wild-type (WT) Lactococcus lactis or L. lactis expressing PilM1, PilM2, PilM4, PilM6 or PilM28 in 20-lL doses.Control larvae were administered with equal volumes of phosphate-buffered saline (PBS).Subsequent to injection, each larva was housed in individual wells of a 12-well plate and incubated at 37°C without feed and monitored over 5 days.Day 5 (a) mean percentage survival and (b) mean health index scoring (using a previously established health index score) calculated for each larvae treatment group are shown.

Figure 6 .
Figure 6.Group A Streptococcus pili stimulate an increased rate of bacterial clearance by G. mellonella.Six wax moth larvae per group were injected with 1 9 10 7 colony-forming units (CFU) of wild-type (WT) Lactococcus lactis or L. lactis gain-of-function strains expressing PilM1, PilM2, PilM4, PilM6 and PilM28.Larvae from each group were homogenized and plated on GM17 agar plates for colony enumeration at 0, 24, 48 and 72 h following inoculation, and CFUs per mL of homogenate were determined for each group.Three independent experiments were performed and representative data from one of these biological repeats are shown as mean AE standard deviation.Statistical significance was determined by two-way ANOVA.P-values were calculated by Holm-S ıd ak's multiple comparisons test and are shown in the box.

Table 1 .
Mean survival rates and health index scores (scale of 1-10) from 10 G. mellonella larvae per dosage following inoculation with different Lactococcus lactis strains on day 5.