DMEM and RPMI 1640 containing stable glutamine were obtained from Biochrom (Berlin, Germany). FCS was purchased from Gibco/Invitrogen (Karlsruhe, Germany); Lipofectamine 2000 and TRIzol from Invitrogen Life Technologies (Darmstadt, Germany); Ultrapure LPS from Salmonella minnesota was supplied by U. Seydel (Forschungszentrum Borstel, Germany); Pam₃CSK₄ and R848 were purchased from Invivogen (San Diego, CA) and CpG1668 was custom synthesized by Eurofins (Luxemburg).

Wildtype (WT), Unc93b1 3d, Tlr23479^−/−, and Tlr13^−/− mice on a C57BL/6 background were housed in the animal facility of the University of Heidelberg under specific pathogen free conditions. Tlr23479^−/− and Tlr13^−/− mice have been described previously (11, 29) and were kindly provided by C. Kirschning (University Hospital Essen, Essen, Germany). Unc93b1 3d mice harboring a H412R missense mutation leading to a non-functional UNC93b1 protein (14) were generously provided by Prof. Dr. M. Freudenberg (Max Planck Institute of Immunobiology and Epigenetics, Freiburg, Germany). All knockout and mutant mice were backcrossed onto the C57BL/6 background for at least 8 generations.

Bone marrow-derived macrophages (BMDMs) and bone marrow-derived myeloid dendritic cells (BMDCs) were produced as described previously (30). For generation of BMDCs, 8 × 10⁶ bone marrow cells were seeded into a 15 cm cell culture plate in differentiation medium (RPMI 1640, supplemented with 10% FCS, 1% penicillin/streptomycin, 0.05 mM 2-mercaptoethanol as well as 1% granulocyte-macrophage colony-stimulating factor (GM-CSF)-containing supernatant from murine GM-CSF-transfected X63 cells (31, 32). Differentiated BMDCs were harvested on day 8. For generation of BMDMs, bone marrow cells were seeded into 15 cm petri dishes and grown in DMEM supplemented with 10% FCS, 1% penicillin/streptomycin and 30–50% L929-supernatant for 7 days. For isolation of neutrophils from mouse bone marrow, a negative selection of neutrophils using immunomagnetic cell separation was performed using an autoMACS pro Separator according to the manufacturer‘s instructions with the neutrophil isolation kit (Miltenyi Biotec, Bergisch Gladbach, Germany). A purity of CD11b/Ly6G double-positive cells of >95% was confirmed by FACS analysis.

The following microbial strains were used for experiments: S. pyogenes ATCC12344, S. pyogenes M49, S. pyogenes M1T1, S. pyogenes AP1 (the latter three strains were a kind gift from B. Kreikemeyer, University Hospital Rostock, Germany), and E. coli ATCC25922. S. pyogenes strains were grown either in Todd-Hewitt-Bouillon supplemented with 0.5% yeast extract (THB; Sigma-Aldrich, Taufkirchen, Germany) or in Brain-Heart-Infusion broth (BHI; Merck, Darmstadt, Germany) at 37°C with 5% CO₂ without agitation in an Erlenmeyer flask. E. coli was grown in Luria-Bertani (LB) medium at 37°C with shaking at 200 rpm. Bacteria were harvested for infection assays within the mid-logarithmic phase growth, centrifuged at 2000 rpm for 10 min and resuspended in PBS or medium to the desired concentration.

For isolation of bacterial RNA (bRNA), S. aureus ATCC25923 was grown in LB medium at 37°C with agitation until mid-logarithmic phase was reached. After lysis with lysozyme (40 mg/ml) for 30 min at 37°C, total bacterial RNA was extracted using TRIzol reagent according to the manufacturer‘s protocol. The isolated RNA was further cleaned up using the RNeasy mini kit (Qiagen, Hilden, Germany) with an integrated on-column DNA digestion step. RNA purity was confirmed by measuring the 260/230 and 260/280 absorbance ratio on a NanoDrop (Thermo Fisher Scientific, Waltham, USA).

Murine cells were generally infected or stimulated at a density of 1.5 × 10⁵ host immune cells/well in a 96-well flat-bottom plate. Only for ROS measurements, cell density was reduced to 5 × 10⁴ cells/well. BMDCs and neutrophils were seeded in antibiotic-free RPMI 1640 supplemented with 10% FCS, whereas BMDMs were seeded in antibiotic-free DMEM supplemented with 10% FCS. Unless indicated otherwise, cells were infected overnight with the indicated bacterial strains at various multiplicities of infection (MOIs), transfected with 2 μg/ml bRNA complexed with Lipofectamine 2000 at a ratio of 1 μl Lipofectamine 2000 per 1 μg RNA or stimulated with Pam₃CSK₄ (1 μg/ml), R848 (1 μg/ml) and LPS (100 ng/ml). For TLR2 blocking experiments, BMDMs were pre-treated with antibodies against mouse TLR2 (Clone T2.5, 5 μg/ml, Biolegend, San Diego, CA) or the appropriate control IgG (Clone MOPC-21, 5 μg/ml, Biolegend, San Diego, CA) 45–60 min prior to infection. 90 min after infection, extracellular bacteria were killed by adding penicillin/streptomycin.

Cell-free supernatants were harvested and analyzed for levels of TNFα, IL-6, and IL-12p40 (BD Biosciences, Heidelberg, Germany) by ELISA according to the manufacturer's instructions. For evaluation of systemic cytokine levels in mice at indicated time points, whole mouse blood was transferred to microtubes with serum gel containing a clotting activator (Sarstedt, Nümbrecht, Germany), left undisturbed for 30 min and then centrifuged for 10 min at 10,000 g to harvest the serum. Serum cytokine profile was determined using either ELISA, the LEGENDplex mouse inflammation panel (Biolegend, San Diego, CA), or a Luminex-based multiplex assay (Bio-Rad, Hercules, USA) according to the manufacturer's instructions.

Under physiological conditions, NO is spontaneously oxidized to its stable and non-volatile breakdown product NO2-, which can be detected by Griess assay. NO2- levels were quantified in cell-free culture supernatants 40 h post infection. Absorbance was measured at 550 nm with 630 nm serving as reference in a photometer (TECAN Sunrise, Männedorf, Switzerland). NO2- concentrations were calculated according to the standard curve obtained by serial dilution of NaNO₂.

For detection of total Reactive oxygen species produced by BMDMs and BMDCs upon infection with S. pyogenes, Total Reactive Oxygen Species Assay Kit (Invitrogen, Waltham, USA) was used. Cells were seeded into sterile black-walled, clear-bottom 96-well flat-bottom plates and pre-treated with ROS Assay Stain stock solution 1 h prior to infection at a final dilution of 1:2000. 24 h post infection, total ROS production was quantified by measuring the fluorescence emission at 520 nm in a microplate reader (BMG FLUOstar Optima, Ortenberg, Germany).

The experimental model of subcutaneous S. pyogenes infection was performed as described previously except for the use of a different strain (33). Briefly, S. pyogenes strain ATCC12344, inoculated from a liquid overnight culture, was grown in Todd-Hewitt-Bouillon supplemented with 0.5% yeast extract at 37°C with 5% CO₂ without agitation. When mid-logarithmic phase growth was reached after 8 h, bacteria were harvested by centrifugation at 2000 rpm for 10 min and resuspended in PBS. The concentration of the bacterial suspension was adjusted to 5 × 10⁶ CFU per 50 μl inoculum using Mc Farland standard, which was verified for each experiment by plating serial 10-fold dilutions of the inoculum on blood agar plates. Age-matched 9- to 11-week-old female WT, Unc93b1 mutant, or Tlr13^−/− mice were lightly anesthetized by i.p. injections of a Ketamine/Xylazine cocktail and subcutaneously infected with 5 × 10⁶ CFU S. pyogenes at the left lower flank after being carefully shaved at the site of infection. 4, 8, and 24 h after infection, mice were sacrificed by CO₂ asphyxiation. Blood was taken by left ventricle puncture and cytokine levels in the serum were analyzed. The lesion of each mouse was excised, and one half was transferred into 4% paraformaldehyde for histopathological examination, the other half was shock-frozen in liquid nitrogen for protein extraction and subsequent cytokine measurements. Mice experiments were approved by the local authorities.

After overnight fixation in 4% buffered paraformaldehyde, representative specimens of the skin were routinely dehydrated, embedded in paraffin, and then cut into 4 μm-thick sections. Sections were stained with H&E according to standard protocols and analyzed using a Leica DM2000 microscope (Wetzlar, Germany). Tissue sections from each animal were then graded for inflammation (score ranging from 0 = no inflammatory cell infiltration to 4 = severe inflammatory cell infiltration), bacterial abundance (score ranging from 0 = no bacteria to 3 = numerous bacteria) and for inflammation depth (score ranging from 0 to 5, with one point each for infiltration of epidermis, dermis, subcutis, superficial musculature, and deep musculature) by a blinded veterinary pathologist.

For analysis of lesional cytokine production, total lesional protein was extracted from tissue samples using 1 mL Tissue Extraction Reagent I (Thermo Fisher Scientific, Waltham, USA) per 100 mg of tissue sample. Lesional cytokine profile was then determined using the LEGENDplex mouse inflammation panel (Biolegend, San Diego, CA) according to the manufacturer's instructions.

To assess statistical significance between multiple groups, the non-parametric Kruskal-Wallis test including Dunn's test to correct for multiple comparisons was used. When normal distribution could be assumed, an ANOVA test was performed, followed by Tukey's post-hoc test. To assess differences in presence or absence of a TLR2 blocking antibody, a linear mixed model with random effect for individual mice was used to account for individual differences between mice, followed by Tukey test. In case of comparisons between two groups, either a two-tailed Mann-Whitney-U-test or, in case of normal distribution, the two-tailed Student's t-test was performed. Significant differences were labeled as follows: ^*p < 0.05; ^**p < 0.01; ^***p < 0.001. Analyses were performed using Graph Pad Prism 7 (Graph Pad Software, San Diego, USA) or using R version 3.3.0 with the packages nlme and multcomp.

This study was carried out in accordance with the recommendations of the German Protection of Animals Act law and approved by the responsible regional authorities (council Karlsruhe, reference number G-98/14 and T-95/15). The protocol was likewise approved by the same council.

Streptococcus pyogenes has previously been reported to engage TLR2 as well as TLR13 on murine BMDMs and BMDCs, yet the relative contribution of both receptors to the overall immune response against this pathogen remains unclear and the literature is conflicting (13, 21). Moreover, limited knowledge is available about a possible inter-strain variability with regard to the relevance of nucleic acid detection for innate immune activation by S. pyogenes. To address these questions, BMDMs from WT, Unc93b1 3d mutant, and Tlr13^−/− mice were infected overnight at various MOIs with different strains of S. pyogenes that are commonly used in the literature. Bacteria were phagocytosed equally in WT and mutant cells (Supplementary Figure 1) and all strains triggered a dose-dependent cytokine secretion in WT BMDMs. However, in Unc93b1 mutant, and Tlr13^−/− BMDMs production of IL-6 and TNFα upon infection with S. pyogenes ATCC12344 and M49 was almost abrogated at lower MOIs (Figures 1A,B and Supplementary Figure 2), indicating a non-redundant role of RNA/TLR13 recognition at low bacterial burden. Of note, nucleic acid dependent recognition was less pronounced at higher MOIs (Figures 1A,B and Supplementary Figure 2), arguing for additional recognition mechanisms that only contribute at high bacterial inoculum. Similar results were observed when bacterial uptake was blocked by cytochalasin D (Supplementary Figure 1). By contrast, S. pyogenes strain AP1 and M1T1 induced relevant cytokine secretion only at higher MOIs and dependency on UNC93B1 or TLR13 was less pronounced (Figures 1C,D). Together, these data reveal a significant inter-strain variability of S. pyogenes in terms of cytokine induction and relevance of nucleic acid sensing for mounting an innate immune response. Strikingly, culture conditions of S. pyogenes did not have an impact on the importance of nucleic acid detection in BMDMs (Supplementary Figure 3). Of note, enhanced IL-6 secretion in Unc93b1 mutated vs. Tlr13-deficient BMDMs was observed after infection with MOI 50 (Figures 1A,C,D), potentially by compensatory upregulation of TLR2 following its stimulation to counterbalance the loss of all endosomal TLRs under high bacterial load (Supplementary Figure 4). A similar trend occurred upon stimulation with Pam₃CSK₄, yet did not reach statistical significance (Figure 1E). In contrast to IL-6, S. pyogenes ATCC12344-induced IL-12p40 production was efficiently abolished at all MOIs in Unc93b1 3d BMDMs, whereas the impairment of IL-12p40 production was less prominent in Tlr13^−/− cells (Figure 1F). These data suggest that endosomal TLRs other than TLR13 are likewise critical for IL-12p40 induction in BMDMs upon S. pyogenes infection, whereas the IL-6 response is mostly dependent on TLR13. Together, these results clearly emphasize the relevance of nucleic acid recognition, especially of bacterial RNA, for the initiation of innate immune responses against S. pyogenes in BMDMs, with differences in their relative contribution to overall activation according to strain, bacterial load, and individual cytokine investigated.

Based on previous work by Fieber et al. (21) and the observation of an upregulation of TLR2 following its stimulation (Supplementary Figure S4), we assumed that TLR2 signaling may be responsible for the residual or even enhanced cytokine production in Tlr13^−/− and Unc93b1 mutant BMDMs at higher MOIs. To verify this hypothesis, WT, Unc93b1 3d, and Tlr13^−/− BMDMs were infected with high MOIs of S. pyogenes ATCC12344 and AP1 (MOI 50 and MOI 25) in the presence or absence of a TLR2-blocking antibody. Inhibition of TLR2 in WT BMDMs only slightly attenuated IL-6 production upon infection with S. pyogenes ATCC12344, whereas cytokine production was completely abolished in Unc93b1 mutant cells and strongly attenuated in Tlr13^−/− cells upon additional blocking of TLR2 (Figure 2A). Thus, these data suggest a redundant role of TLR2 in sensing S. pyogenes that becomes visible only under conditions of high bacterial load and impeded endosomal TLR signaling, with TLR13 playing the most prominent role among all nucleic acid sensing TLRs. Similar results were obtained when BMDMs were infected with the strain AP1 (Supplementary Figure 5). The presence of a TLR2-blocking antibody efficiently inhibited responses to the TLR2 ligand Pam₃CSK₄ without interfering with TLR13 and TLR4 signaling or the general responsiveness of BMDM as demonstrated by using isotype control antibodies (Figures 2B,C), thereby confirming efficacy and specificity of the antibody.

To further dissect the relative contribution and potential redundancies of TLR2 and endosomal TLRs with respect to S. pyogenes sensing, the response of cells with a combined deficiency in TLR 2, 3, 4, 7, and 9 was investigated. Intriguingly, neither IL-6 secretion nor IL-12p40 production were impaired in Tlr23479^−/− BMDMs (Figures 2D,E), indicating a redundant function of these receptors in recognition of S. pyogenes even at high MOIs. The same was observed when cells were infected with S. pyogenes AP1 strain (Supplementary Figure 5). Taken together, the results presented so far corroborate the major role of TLR13 and TLR2 in recognition of S. pyogenes by BMDMs. Moreover, the data identify the TLR2 and TLR13 pathway as being redundant at higher MOIs, whereas TLR13 signaling is shown to play an essential, non-redundant role at lower MOIs and to compensate completely for the loss of TLR2 and other endosomal TLRs in the defense of S. pyogenes. Thus, bacterial RNA represents the dominant immune stimulus for mounting an innate immune response against S. pyogenes in BMDMs.

As host defense mechanisms against pathogens physiologically engage several cell types, we next examined whether S. pyogenes RNA also represents the dominant immune stimulus in cell types other than macrophages. For this purpose, we measured cytokine production in bone marrow-derived dendritic cells (BMDCs) from WT, Unc93b1 3d mutant, and Tlr13^−/− mice infected overnight with various MOIs of S. pyogenes ATCC12344. In line with the data obtained in macrophages, both IL-6 and IL-12p40 secretion in BMDCs were critically dependent on endosomal TLR signaling, especially at lower MOIs (Figures 3A,B), thus further substantiating the critical role of nucleic acid detection in cytokine induction. However, in striking contrast to macrophages, BMDCs from Tlr13^−/− mice displayed only slightly decreased cytokine levels compared to their WT counterparts (Figures 3A,B), suggesting that in BMDCs, non-redundant recognition of S. pyogenes RNA by TLR13 was less relevant for the induction of an immune response. Yet, BMDCs deficient in TLR2/3/4/7/9 produced comparable amounts IL-6 (Figure 3C) and IL-12p40 (Supplementary Figure 6) as WT cells, implying that in the absence of TLR2/3/4/7/9 signaling the TLR13 pathway becomes more prominent compared to WT cells.

Neutrophils represent the first line of defense in bacterial infections (22), but to date, it has not yet been investigated which TLRs are involved in recognition of S. pyogenes in neutrophils. We therefore further investigated the dependency of murine neutrophils on nucleic acid detection in defense of S. pyogenes. Infection of neutrophils isolated from WT, Unc93b1 3d, and Tlr13^−/− mice with S. pyogenes ATCC12344 revealed a similar stimulation pattern as observed in BMDCs: While in neutrophils lacking functional UNC93b1, TNFα production triggered by S. pyogenes was strongly impaired and IL-12p40 secretion was even completely abolished, the secretion of both cytokines was less attenuated in Tlr13^−/− cells, especially in case of IL-12p40 (Figures 3D,E). This finding clearly demonstrates the pivotal role of nucleic acid detection for mounting an immune response against S. pyogenes in neutrophils, whereas specific recognition of S. pyogenes RNA via TLR13 appears to be less prominent. In line with previously published data (34), we observed that murine neutrophils only barely respond to single TLR ligands including purified bacterial RNA (Figures 3D,E). Yet, neutrophils from all genotypes responded equally to infection with live E. coli which served as positive control (Figure 3F).

In conclusion, our results highlight nucleic acid detection as a prerequisite for an effective immune response in both BMDCs and neutrophils. However, in contrast to macrophages, TLR13, and the endosomal nucleic acid sensing receptors TLR3/7/9 are largely redundant in their function.

Reactive nitrogen species (RNS) and reactive oxygen species (ROS) have been demonstrated to be a major part of an effective host response toward pathogens by exerting a direct antimicrobial activity (35–37) and thus containing the infection (38–40). However, to date, the contribution of endosomal TLR signaling to NO formation or ROS production in S. pyogenes infections has not yet been evaluated. We therefore assessed these parameters in BMDMs and BMDCs from WT, Unc93b1 3d, and Tlr13^−/− mice upon infection with S. pyogenes ATCC12344 in vitro. NO generation was strongly reduced in BMDMs lacking either functional endosomal TLRs or TLR13 alone when compared to their WT counterparts, especially at lower MOIs (Figure 4A). A similar pattern was observed for ROS production although data did not reach statistical significance for Tlr13^−/− cells (Figure 4B). Likewise, S. pyogenes induced NO and ROS formation was also highly dependent on nucleic acid sensing in BMDCs, as demonstrated by an impaired production of both antimicrobial agents in Unc93b1 mutant cells (Figures 4C,D). In line with the cytokine secretion profiles (Figures 1A, 3A,B), the relative contribution of endosomal TLR signaling to NO production vanished with higher MOIs in both cell types, with recognition of S. pyogenes via TLR13 being less prominent in BMDCs as compared to BMDMs. Together, these findings identify nucleic acid sensing as prominent stimulus for NO formation and as contributor to ROS production in BMDMs and BMDCs upon S. pyogenes infection.

Considering the importance of nucleic acid sensing for the induction of innate immune responses in vitro, we next set out to investigate whether nucleic acid detection is also required for defense against S. pyogenes in vivo. WT, Unc93b1 mutant, and Tlr13^−/− mice were therefore subcutaneously infected with S. pyogenes strain ATCC12344 and sacrificed at the indicated time points. This established mouse model of soft tissue infection mimics human skin infections with S. pyogenes culminating in necrotizing fasciitis and sepsis (41). During the early phase of infection, Unc93b1 mutant mice displayed significantly decreased systemic IL-6 levels compared to their WT controls (Figure 5A), indicative of an insufficient recognition of S. pyogenes by the innate immune cells, thus potentially favoring systemic dissemination. However, 24 h post infection, we observed overall increased systemic cytokine and chemokine levels in mice deficient in endosomal TLR signaling (Figures 5A,B), presumably due to an increased bacteremia, as demonstrated by a tendency toward higher bacterial burden in spleens from Unc93b1 mutant mice compared to their WT counterparts (Supplementary Figure 7). At the same time, histopathological examination 24 h post infection revealed more severe lesions in Unc93b1 mutant mice with an elevated bacterial load and more inflammatory cells invading into deeper layers of the lesions (Figures 5C,D). Notably, markedly enhanced lesional levels of the alarmin IL-1α were observed in infected mice compared to non-treated controls (Figure 5D), thus confirming the presence of a relevant tissue damage and necrosis. Together, our data demonstrate a failure in early recognition of S. pyogenes ATCC12344 in Unc93b1 mutant mice, leading to a reduced containment of S. pyogenes in the course of infection with subsequent spread and systemic inflammation. In line with these results, Tlr13^−/− mice also displayed significantly enhanced systemic IL-6 levels compared to their WT controls 24 h after infection, albeit they did not show an overall increased systemic inflammatory response as observed in Unc93b1 mutant mice (Figure 5E). Consistently, Tlr13^−/− mice exhibited a slightly, yet not significantly, enhanced bacterial and inflammatory burden at the site of infection, with a mildly increased local infiltration of inflammatory cells (Figure 5F). This enhanced local inflammation might be due to elevated lesional GM-CSF levels found in Tlr13^−/− mice, promoting the recruitment of inflammatory cells to the infection site (Figure 5F).

In conclusion, our results argue for a crucial contribution of nucleic acid detection via endosomal TLRs to defense against S. pyogenes in vivo.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.