ExPortal and the LiaFSR Regulatory System Coordinate the Response to Cell Membrane Stress in Streptococcus pyogenes

Bacterial two-component systems sense and induce transcriptional changes in response to environmental stressors, including antimicrobials and human antimicrobial peptides. Since the stresses imposed by the host’s defensive responses may act as markers of specific temporal stages of disease progression or host compartments, pathogens often coordinately regulate stress response programs with virulence factor expression. The mechanism by which bacteria recognize these stresses and subsequently induce transcriptional responses remains not well understood. In this study, we showed that LiaFSR senses cell envelope stress through colocalization of LiaF and LiaS with the group A Streptococcus (GAS) ExPortal and is activated in direct response to ExPortal disruption by antimicrobials or human antimicrobial peptides. Our studies shed new light on the sensing of cell envelope stress in Gram-positive bacteria and may contribute to the development of therapies targeting these processes.

clearance (2). In response, many bacterial pathogens have developed a number of strategies to disarm or mitigate the effects of AMPs in order to establish infection (1).
The bacterial membrane is a complex structure critical to maintaining the integrity of the cell. It is now recognized that many bacterial cell processes are maintained in discrete locations within the membrane that are referred to as functional membrane microdomains (FMMs) (3)(4)(5). On the basis of the association with the general secretory pathway (6), the microdomain in group A Streptococcus (GAS) has been termed the "ExPortal" and, in addition to protein secretion and processing, colocalizes with proteins critical in peptidoglycan precursor synthesis (7). Not surprisingly, given the association with essential cell processes such as cell wall synthesis and division, the bacterial FMM is a key target of host defenses. The naturally occurring lipopeptide daptomycin (DAP) targets FMMs of Gram-positive bacteria, displacing membraneassociated proteins essential for cell wall maintenance (8). Similarly, the human alphadefensin human neutrophil peptide 1 (hNP-1), restricted to azurophilic granules of neutrophils (9), preferentially disrupts the ExPortal, inhibiting protein secretion and processing (10,11). The bacterial FMM is an attractive target for the development of novel antimicrobials in the battle against resistance (12). However, much information is lacking regarding FMM content and organization and the downstream effects of antimicrobial-induced FMM disruption.
The ability of bacterial pathogens to colonize a host and grow at the site of infection is dependent upon precise gene regulation in response to external stimuli. To accomplish this feat, many bacteria utilize gene regulatory two-component systems (TCSs) composed of a membrane-bound sensor histidine kinase (HK) that, in response to a stimulus, activates a cognate response regulator (RR) via phosphotransfer (13). Not surprisingly, some bacterial TCSs have been shown to respond to human AMPs (14,15). The LiaFSR TCS, unique to Gram-positive bacteria, is composed of an HK (LiaS), an RR (LiaR), and an accessory membrane protein of unknown function (LiaF). LiaFSR is a critical component of the cell envelope stress response and is activated following exposure to antimicrobials targeting the cell envelope (e.g., bacitracin [BAC] or daptomycin) or AMPs (e.g., LL-37) (16). Interest in the LiaFSR TCS is rooted in the association between antimicrobial resistance and mutations within the individual TCS components or its downstream effectors (17,18). While much is known regarding LiaFSR and the development of resistance to antimicrobials, the external host stimuli, mechanisms of signaling, and downstream effects of LiaFSR activation in pathogenic Gram-positive bacteria are largely unresolved.
The HK (LiaS) of the LiaFSR regulatory system is among the members of a specialized class referred to as intramembrane (IM) sensing HKs (IM-HKs). In contrast to HKs that have large extracytoplasmic sensing domains, IM-HKs possess two or more transmembrane domains that, together with accessory membrane proteins (e.g., LiaF), respond to changes in the cell membrane (19). Most of what is known regarding IM-HK activation and signal transduction has been derived from TCSs that respond to antimicrobialinduced stress (e.g., LiaS and BceS) (16,20). Given the apparent need for accessory proteins and proper IM-HK function, precise subcellular colocalization is assumed but such information for many bacterial pathogens is lacking. Recently, it was shown that the membrane microdomain protein content of Staphylococcus aureus includes IM-HKs such as SaeS and the LiaS homolog VraS (3). High-resolution imaging has also shown subcellular focal interaction between membrane kinases and target regulatory proteins (21,22). Combined, those studies indicate a critical link between membrane microdomains and signaling pathways.
Here, we begin to define the relationship between LiaFSR and the membrane microdomain (ExPortal) in GAS by testing the hypothesis that signal transduction through LiaFSR is directly linked to the GAS ExPortal. We discovered that LiaF and LiaS colocalize with the GAS ExPortal and respond to antimicrobial and human AMPinduced membrane stress. In addition, we have uncovered a critical role of cardiolipin (CL) in LiaS localization and of the accessory protein LiaF in GAS ExPortal integrity.

Antimicrobials disrupt LiaF, LiaS, and ExPortal colocalization and activate
LiaFSR. Previous work has shown that disruption of the ExPortal by AMPs resulted in redistribution of ExPortal-associated proteins such as SecA and HtrA (11). We hypothesized that activation of LiaFSR is triggered by disruption of the ExPortal by antimicrobials or AMPs. Data on LiaFSR from Bacillus subtilis suggest that LiaF may act as an inhibitory protein with respect to LiaS (27). Also, antimicrobials that target the bacterial cell membrane disrupt FMMs such as the GAS ExPortal (11). Thus, we examined whether antimicrobials that disrupt ExPortal organization also affect LiaF and LiaS colocalization and subsequent LiaFSR activation. MGAS10870 was grown to midexponential phase and treated with subinhibitory concentrations of six commonly used antimicrobials. GAS cells were harvested and examined for ExPortal integrity using nonyl acridine orange (NAO) staining and for LiaFSR activation using targeted spxA2 transcript level measurements. No significant differences in growth compared to untreated cells were observed for the concentration of antimicrobials used in the assay (Fig. S5). The typical focal NAO staining pattern of the ExPortal was disrupted by treatment with cell membrane or cell wall-active antimicrobial agents but not by treatment with tetracycline (protein synthesis inhibitor) ( Fig. 2A and B). Treatment also resulted in the concomitant redistribution of LiaF and LiaS from single foci to cells with multiple foci or a diffuse pattern ( Fig. 2A and B), indicating a dependence on ExPortal integrity. Compared to untreated cells, we observed significant increases in spxA2 transcript levels following the addition of cell wall-active antimicrobials (Fig. 2C). In addition, as a more direct measure of LiaFSR activation, we assayed LiaR phosphorylation levels using Phos-tag following bacitracin (BAC) or daptomycin (DAP) treatment. We observed significantly increased LiaR phosphorylation following antimicrobial treatment using cell envelope-active compounds, confirming LiaFSR activation (Fig. 2D). The level of LiaR phosphorylation observed following antimicrobial treatment was similar to that seen with an isogenic mutant lacking LiaF (ΔliaF), confirming the inhibitory role of LiaF in the LiaFSR system. Further, activation was specific to cell envelope-active antimicrobials and did not represent a general stress response as no activation was seen with tetracycline-similarly to strains lacking the HK (ΔliaS) or mutated at the essential aspartate residue of LiaR (LiaR-D56A) (Fig. 2D). Treatment with antimicrobials did not result in LiaF or LiaS degradation (Fig. 2E). Overall, the data show that disruption of the ExPortal results in the redistribution of LiaF and LiaS leading to activation and are consistent with a model where LiaF acts as an inhibitor of LiaS activity.
hNP-1 disrupts LiaF and LiaS colocalization and activates LiaFSR 3CS. Previously, it was shown that human AMPs affect ExPortal integrity (11). Thus, we sought to examine the effect of the human AMPs hNP-1 (alpha-defensin), hBD-1 (beta-defensin), and LL-37 (cathelicidin) on LiaFSR activation. We first evaluated the effect of the AMPs on GAS growth in vitro. Only the alpha-defensin hNP-1 demonstrated growth inhibition (Fig. S6). NAO staining of cells following exposure to individual AMPs at the maximal HtrA-GFP were subjected to immunogold electron microscopy and stained to detect GFP (12-nm-diameter beads) and FLAG (18-nm-diameter beads). (I) Immunoblot to detect LiaS and LiaF in the ExPortal membrane fractions (DRF). Membrane fractions were prepared as described in Materials and Methods. LiaS-GFP, LiaF-GFP, YajC-GFP, and HtrA-GFP were identified by Western blotting using an anti-GFP antibody. DSF, detergent-sensitive fraction; DRF, detergent-resistant fraction; All, total membrane fractions.
LiaFSR-ExPortal Codependence in GAS ® amount that did not impact growth (Fig. S6) revealed ExPortal disruption by hNP-1 but not by hBD-1 or LL-37 ( Fig. 3A) and was similar to that observed for antimicrobials ( Fig. 2A and B). In vitro exposure to hNP-1 resulted in the redistribution of LiaF and LiaS, resulting in multiple foci or a diffuse pattern ( Fig. 3B and C). As seen following antimicrobial treatment, exposure to human AMPs did not alter LiaF or LiaS protein levels (Fig. 3D). LiaFSR activation, as assessed by spxA2 transcript level, was observed only following the addition of hNP-1 (Fig. 3E). No significant changes in spxA2 transcript levels were observed in the presence of hBD-1 or LL-37 (Fig. 3E). Importantly, the concentration at which hNP-1 inhibition was observed (Ͼ50 g/ml) was similar to that The gene-transcript levels of spxA2 were quantitated (relative to an endogenous control, tufA) following treatment using the indicated antimicrobial as described for panels A and B and compared to those quantitated for the untreated sample. ns, not significant. (D) Determination of LiaR phosphorylation (LiaRϳP) level using Phos-tag gel. Samples were harvested at a time identical to the assessments described for panels A to D. LiaRϳP levels (percentage relative to the WT, untreated) are indicated below the individual lanes. The LiaR-D56A strain was unable to be phosphorylated due to a mutation of the conserved/ essential aspartate residue. Similarly, LiaR exists in an unphosphorylated state in the absence of LiaS (ΔliaS). ND, not done. (E) The effect of antimicrobials on the LiaF and LiaS cellular protein levels. After exposure to antimicrobial agents as described for panel A, LiaF-GFP and LiaS-GFP were monitored by Western blotting using an anti-GFP antibody. P values were determined by Student's t test. ***, P Ͻ 0.001 relative to untreated. Abbreviations: BAC, bacitracin; CST, colistin; DAP, daptomycin; NIS, nisin; PMB, polymyxin B; TET, tetracycline; VAN, vancomycin.
Lin et al. ® expected in purulent fluid (28). Finally, we assessed the LiaR phosphorylation level following hNP-1 exposure and observed significantly increased phosphorylation compared to controls (untreated and phosphorylation-deficient LiaR mutant) (Fig. 3F). Thus, our data confirm ExPortal sensitivity to the human AMP hNP-1 and support the idea of LiaFSR activation as a result of ExPortal disruption and LiaF/LiaS redistribution in the cell membrane.
Elimination of membrane cardiolipin results in the loss of LiaS/ExPortal colocalization. Cardiolipin (CL) is an anionic phospholipid known to contribute to mem-  Fig. 2A and B. (D) The effect of human AMPs on LiaF and LiaS cellular protein levels. Following exposure to human AMPs as described for panels A to C, LiaF-GFP and LiaS-FLAG were monitored by Western blotting using an anti-GFP antibody and anti-FLAG antibody. (E) The activation of LiaFSR 3CS in the presence of hNP-1, hBD-1, and LL-37 was determined by qRT-PCR analysis for spxA2 as described in Materials and Methods. Data are shown as means Ϯ SD of results from three independent experiments. (F) Determination of LiaR phosphorylation (LiaRϳP) levels using Phos-tag gel. LiaRϳP levels (percent relative to WT, untreated) are indicated below the individual lanes. P values were determined by Student's t test. ***, P Ͻ 0.001 relative to untreated. ns, not significant.
LiaFSR-ExPortal Codependence in GAS ® brane protein stability and localization (29). Phosphatidylglycerol (PG) may be converted to CL by cardiolipin synthase (CLS), of which GAS genomes possess a single copy (30). Deletion of cls (Δcls) in GAS results in complete loss of membrane CL, but the ExPortal remains intact, indicating that additional factors contribute to maintenance of the membrane microdomain (30). We hypothesized that LiaF and/or LiaS association with the ExPortal was CL dependent. We tested our hypothesis by constructing an isogenic mutant lacking CLS (Δcls, Table S1) that was subsequently used to introduce GFP-and FLAG-tagged LiaF and LiaS proteins, respectively, as shown in Fig. 1. The mutant lacking CLS (Δcls) showed a focal staining pattern similar to that seen with the parental strain when stained with NAO, indicating an intact ExPortal (Fig. 4A). The Δcls strain contained high levels of phosphorylated LiaR consistent with LiaFSR activation (Fig. 4B). In contrast, a strain lacking the HK (ΔliaS) or mutated at the essential aspartate (D) residue in LiaR (LiaR-D56A) showed WT levels of LiaR phosphorylation (Fig. 4B). Similarly to the dissociation of LiaF and LiaS and resultant LiaFSR activation observed  (Fig. 4C). Immunofluorescence microscopy showed that the Δcls strain contained single LiaF foci but had multiple points for LiaS (Fig. 4D) that did not overlap (Fig. 4F). LiaF also demonstrated colocalization with HtrA in the Δcls strain, verifying an association with the ExPortal (Fig. S7). Consistent with a cardiolipin-specific effect, complementation of the Δcls mutant in trans (Table S1) showed a restoration of single LiaF and LiaS foci with colocalization by IF microscopy (Fig. 4E and F). In addition, we observed significantly decreased LiaR phosphorylation upon complementation of the Δcls mutant that was similar to that seen with the WT (Fig. 4G). The data indicate that cardiolipin is critical for LiaS association with the GAS ExPortal, thereby maintaining LiaFSR activation in the basal state.
Mutants lacking LiaF exhibit altered ExPortal integrity. Membrane microdomain stability may also be driven through specific protein-protein interaction (3). Thus, we next hypothesized that LiaF and/or LiaS may contribute to the maintenance of the GAS ExPortal. We generated isogenic deletion mutants lacking LiaF (ΔliaF) or LiaS (ΔliaS) in the parental strain and performed NAO staining to visualize the ExPortal. The mutant lacking LiaS (ΔliaS) showed a focal staining pattern similar to the parental strain pattern (Fig. 5A and B). However, in contrast to that observed for the WT and ΔliaS strains, the ΔliaF strain displayed primarily septal and circumferential cell staining that was restored upon complementation in trans (Fig. 5A and B). We also compared the NAO staining patterns of the ΔliaF and ΔliaS strains to that of a mutant lacking the response regulator (ΔliaR)-a cytoplasmic protein with no known membrane association-and observed a WT staining pattern in the ΔliaR strain ( Fig. 5A and B). Complementation of the ΔliaF mutant also restored WT levels of LiaR phosphorylation (Fig. 5C). Consistent with disruption and the requirement for an intact ExPortal for SpeB (cysteine protease) secretion (10), we observed nearly absent SpeB activity in the ΔliaF mutant but not the ΔliaS mutant (Fig. 5D) similar to that seen with a mutant lacking the essential regulator for speB transcription the (ΔropB mutant; Fig. 5D) (31). In total, the data derived from levels in the GAS WT, the ΔliaF mutant, and a ΔliaF mutant strain carrying pLZ12Km2::LiaF using Phos-tag gel. LiaRϳP levels (% relative to WT) are indicated below the individual lanes. (D) SpeB activity determined using FITC-casein assay as described in Materials and Methods. Shown is SpeB activity from culture supernatants relative to the parental, WT strain from three biological replicates (means and SD). The isogenic ΔropB deletion strain (regulator of SpeB) was included as a negative control. P values were determined by Student's t test. ***, P Ͻ 0.001 relative to WT.

DISCUSSION
Sensing of environment is critical to bacterial pathogenesis. Unique among TCSs are intramembrane-sensing histidine kinases as they lack external sensing domains and instead rely upon changes within the cell membrane and accessory proteins to trigger the cascade of events leading to a transcriptional response (16). LiaFSR, found only in Gram-positive bacteria, is one such system and has garnered much interest due to its role in the development of antibiotic resistance (17,(32)(33)(34). The current investigation provides evidence that the histidine kinase, LiaS, and the accessory protein, LiaF, of the human pathogen Streptococcus pyogenes reside within a discrete membrane microdomain, the ExPortal. In response to antimicrobial-or AMP-induced ExPortal stress, LiaFSR acts as a critical signal transducer to coordinate the bacterial response (Fig. 6). Surprisingly, the data support the essentiality of LiaF for membrane microdomain integrity in GAS. Combined with data indicating that LiaF (YvqF/VraT) and LiaS (VraS) homologs colocalize with membrane microdomains in S. aureus (3), the current report provides a model applicable to many Gram-positive pathogens.
In addition to the association with the GAS ExPortal, our data suggest distinct mechanisms by which LiaF and LiaS localize within the cell membrane and potential roles in microdomain formation. Two primary mechanisms are attributed to protein focalization and membrane microdomain formation: differences in lipid content and  .g., spxA2). LiaS most likely exists as a dimer but for simplicity is shown as a monomer. The LiaF/LiaS ratio in GAS is unknown but, based on studies in Bacillus (38), may be present in relative excess. For simplicity, the LiaF/LiaS ratio is represented as 1:1.
protein-protein interaction (35). Constituent lipids of bacterial microdomains include carotenoids and cardiolipin, which are critical for microdomain formation in Staphylococcus aureus (3) and in Escherichia coli and Bacillus (36,37), respectively. Our data are consistent with this concept in that the LiaS protein mislocalizes in a strain lacking the only known GAS cardiolipin synthase, suggesting a critical role of cardiolipin in LiaS association with the ExPortal. It is now recognized that bacterial flotillins are critical for protein aggregation at membrane microdomains promoting protein-protein interactions (3). That a GAS strain lacking LiaF exhibits complete ExPortal disruption suggests a possible role of LiaF in maintaining microdomain assembly. In Bacillus, LiaF is in relative excess compared to LiaS (estimated 14:1 ratio) (38). Thus, if also present in relatively large amounts in the GAS ExPortal, LiaF may promote the membrane lipid and protein aggregation essential for microdomain formation. Alternatively, given that the absence of LiaF results in constitutive activation of LiaS and LiaR, changes in gene transcription resulting in altered cell membrane lipid composition may affect microdomain assembly. Further work is needed to determine the role of LiaFSR in microdomain formation in GAS and other Gram-positive bacteria.
The finding that inhibitors of cell wall synthesis such as vancomycin disrupt the GAS ExPortal is of particular interest in the climate of antimicrobial resistance. Disassembly of the GAS ExPortal is likely related to the key association of ExPortal integrity and an intact peptidoglycan layer (7). The GAS organism remains quite susceptible to betalactam antibiotics despite recent reports of penicillin-binding protein mutations resulting in modestly increased MICs (39). However, studies in methicillin-resistant Staphylococcus aureus (MRSA) showed that targeting the FMM leads to a resensitization to beta-lactams, primarily through disruption of PBP2a oligomerization (3,40). This socalled seesaw effect is exploited to treat recalcitrant MRSA infections (41). Together, these findings further imply a common theme among Gram-positive pathogens that can be exploited to combat resistance.
Our data indicate a refined response by GAS to human AMPs. GAS is a humanspecific pathogen, and centuries of coevolution have resulted in specificity of GAS proteins for human molecules. Previously, it was demonstrated that the human cathelicidin LL-37 induces CovRS-mediated virulence gene expression (42). Subsequently, the same group showed that a specific fragment of LL-37 interacts with the extracellular domain of CovS, resulting in altered virulence gene expression (15). Our data indicate that, consistent with activation of intramembrane-sensing TCSs (16), LiaFSR becomes activated upon membrane perturbation (assessed by ExPortal/FMM integrity) by a specific human AMP, in this case hNP-1. Given that the primary reservoir of hNP-1 in the human host is the neutrophil (43), the effect of hNP-1 on the ExPortal and LiaFSR activation (relative to that seen with hBD-1 or LL-37) highlights the importance of the GAS-neutrophil interaction in GAS infections.
In summary, we describe a heretofore-unrecognized mechanism by which GAS responds to the human host by linking specific membrane perturbations (FMM) with TCS activation (LiaFSR). The critical role of the membrane microdomains in bacterial physiology and the high degree of LiaFSR conservation suggest a common theme among Gram-positive bacterial pathogens.
LiaFSR-ExPortal Codependence in GAS ® Generation of isogenic mutants in MGAS10870. The plasmids and primers used for isogenic mutant generation are listed in Tables S2 and S3, respectively. Gene fusions to generate GFP-and FLAG-labeled proteins and in-frame gene deletions were constructed using overlap sewing PCR (45). Strains expressing markerless superfolded green fluorescent protein (sfGFP) or 3xFLAG (FLAG) (Table S1) were generated using the allelic exchange plasmid pJL1055 as described previously (46). The in-frame deletions of liaF, liaS, liaR, cls, and ropB were made by replacement with either a kanamycin resistance gene (aph or ropB; Table S1) or a spectinomycin resistance gene (aad9) in the MGAS10870 parental strain. All isogenic mutant strains underwent whole-genome sequencing for confirmation and lacked spurious mutations.
Challenge with antimicrobials and AMPs. Unless otherwise indicated, bacteria were cultured overnight in liquid THY medium, diluted 1:100 in fresh, prewarmed THY medium, and then cultured to the early logarithmic phase of growth (optical density at 600 nm [OD 600 ] of 0.4). The antimicrobial or AMP was then added to the final concentration indicated, and incubation continued for an additional hour. Untreated control cultures were grown in parallel. At selected time points following challenge with antimicrobials or AMPs, aliquots were removed from cultures. LiaFSR 3CS activity was monitored by the use of quantitative real-time PCR (qRT-PCR) or Pho-tag gels as described below. Protein distribution and ExPortal were visualized by immunofluorescence or NAO staining as described below. Viability was assessed by determination of CFU following brief vortex mixing for 5 min to disrupt streptococcal chains, serial dilution in phosphate-buffered saline (PBS), and plating on SBA. Data reported represent means and standard errors of the means derived from a minimum of three independent experiments.
Dansyl-polymyxin B synthesis. Dansyl-polymyxin B (dPM) was synthesized as described previously (47) with minor modifications. Briefly, polymyxin B sulfate (Sigma-Aldrich) was dissolved in 0.1 M NaHCO 3 to reach a final concentration of 26 mM. Dansyl chloride (Acros Organics, Thermo Fisher Scientific, Waltham, MA) was dissolved in acetone to reach a final concentration of 46 mM. Polymyxin B was mixed with dansyl chloride at a molecular ratio of 3:2, and the mixture was placed in the dark for 120 min at 37°C. After incubation, the dPM and the unreacted dansyl chloride were separated by passage through a Sephadex G-25 column (Sigma-Aldrich). The fractions containing the dPM were extracted into n-butanol. The pure dPM was dissolved in the buffer of 5 mM HEPES (pH 7.0) and stored in aliquots at Ϫ20°C. The concentration of the dPM was determined by dinitrophenylation assay (48).
Cellular staining and fluorescence microscopy. The location and integrity of a membrane microdomain enriched in anionic phospholipids were assessed by staining with 10᎑nonyl acridine orange (NAO; Invitrogen) (30) or dPM (11). Briefly, strains were grown to an OD 600 of 0.4 and NAO or dPM was added to reach a final concentration of 1 M or 10 M, respectively. Cells were incubated an additional 30 min at 37ЈϾò/ϾC and were harvested by centrifugation. After three washes with PBS, cells were attached to glass slides pretreated with poly-L-lysine. Samples were examined using a Keyence model BZ-X700 all-in-one fluorescence microscope and images captured using BZ-X Analyzer software (Keyence, Itasca, IL). Where indicated, simultaneous treatment with two reagents was conducted in order to assess colocalization of staining. In such experiments, GAS cell walls were visualized by staining with wheat germ agglutinin (WGA) Alexa Fluor 350 conjugate (5 g/ml; Invitrogen, Thermo Fisher). Focal localization in images of cells was quantified by scoring staining at a unique (single) focus, at multiple foci, or nonspecifically (the staining was of homogeneous intensity around the cellular circumference). Data presented for each condition represent means and standard errors of the means of results derived from at least three independent experiments and examination of a minimum of 300 stained cells as indicated in the text. Immunofluorescent microscopy. Aliquots from untreated or antimicrobial/AMP-treated cultures (see above) were collected as indicated and centrifuged to pellet cells. For fixation, cells were resuspended in 100 mM phosphate buffer (pH 7.4) containing 4% paraformaldehyde (Avantor, VWR, Radnor, PA) and 0.5% glutaraldehyde (Fisher Scientific) and incubated for 1 h at 4°C. After fixing, cells were washed three times with 0.1 M phosphate buffer (pH 7.4) and attached to poly-L-lysine-coated glass slides. To increase the permeabilization of cells, the fixed cells were treated with ice-cold methanol following the treatment of PlyC lysin (a generous gift from Daniel C. Nelson, University of Maryland, USA) (49). Antisera used included a polyclonal rabbit anti᎑GFP antiserum (Abcam, Cambridge, MA) used at a dilution of 1:200 that was detected using Alexa Fluor 488᎑labeled goat anti᎑rabbit IgG (Abcam) at 1:500 and a polyclonal mouse anti᎑FLAG antiserum (Invitrogen, Thermo Fisher) used at a dilution of 1:200 that was detected using Alexa Fluor 647᎑labeled goat anti᎑mouse IgG (Abcam) at 1:500. Slides were mounted in an antifade medium (Vector Laboratories, Burlingame, CA) and images captured and staining patterns quantified as described for NAO staining above.
Immunogold labeling and transmission electron microscopy. Cells were prepared and fixed as described above. Samples were then dehydrated and infiltrated with LR White resin (Electron Microscopy Sciences, Hatfield, PA) as follows: 50% ethanol for 15 min, 70% ethanol for 15 min, 80% ethanol for 10 min, LR White resin/ethanol (2:1 LR White resin to 70% [vol/vol] ethanol) for 60 min, 100% LR White resin for 60 min, 100% LR White resin overnight, 100% LR White resin for 30 min, and 100% LR White resin for 30 min. Samples were then embedded in fresh LR White resin and polymerized at 55°C for 72 h. Samples were sectioned with a Leica Ultracut UCT ultramicrotome in 70-to-80-nm sections and immunogold labeled with a polyclonal rabbit anti᎑GFP antiserum (Abcam) or a polyclonal mouse anti᎑FLAG antiserum (Invitrogen, Thermo Fisher) followed by 12-nm-diameter colloidal gold-conjugated goat anti-rabbit IgG (Abcam) or 18-nm-diameter colloidal gold-conjugated donkey anti-mouse IgG (Abcam). Sections were stained with 1% uranyl acetate and 0.01% lead citrate for 10 and 5 min, respectively, and viewed on a JEOL JEM 1400 120-kV transmission electron microscope. Double immunolabeling was performed using the two primary antibodies in tandem followed by 18-nm-diameter colloidal gold-conjugated donkey anti-mouse IgG and 12-nm-diameter colloidal gold-conjugated goat anti-rabbit IgG. All labeling experiments were processed in parallel omitting one primary antibody. These controls were consistently negative for the corresponding secondary gold-conjugated antibody (data not shown).
GAS membrane ExPortal isolation. GAS cultures were prepared as described above. Pellets were resuspended in PBS buffer supplemented with 1 mM phenylmethylsulfonyl fluoride (PMSF), 20 g/ml DNase I (Sigma), and 5 g/ml PlyC lysin and incubated for 30 min at 25°C. Cell suspensions were disrupted using a French press. Unbroken cells and debris were removed by centrifugation (10 min, 16,000 ϫ g, 4°C). Membrane fractions were collected by ultracentrifugation (1 h, 100,000 ϫ g). For ExPortal isolation, the membrane fractions were processed using a CelLytic MEM protein extraction kit (Sigma). The detergent-resistant fraction (DRF) and detergent-sensitive fraction (DSF) were separated according to the manufacturer's protocol.
Extraction of RNA for quantitative real-time PCR. Samples from GAS cultures grown for NAO staining or immunofluorescence were harvested by centrifugation, and RNA was isolated and purified with a RNeasy minikit (Qiagen, Germantown, MD) according to manufacturer protocol with modifications for Gram-positive bacteria. The quantity and quality of RNA were determined using a NanoDrop spectrophotometer (Thermo Fisher) and an Agilent 4200 TapeStation system (Agilent Technologies, Santa Clara, CA). TaqMan (Applied Biosystems, Thermo Fisher) quantitative real-time PCR (qRT-PCR) analysis of cDNA produced using SuperScript III reverse transcriptase (Invitrogen, Thermo Fisher) was performed with an CFX96 Touch real-time PCR detection system (Bio-Rad Laboratories, Hercules, CA) according to the manufacturer's instructions, with 50 ng total RNA and a 10 nM concentration of each of the primers (forward and reverse) in a 20-l volume. Real-time PCR primers and probes used in the analyses are listed in Table S3. Transcript levels of genes were calculated by relative quantification using the threshold cycle (ΔΔC T ) method as described previously (50), with a tufA internal reference gene used as the normalizing gene (51). All reactions were performed in technical triplicate using RNA purified from at least three biological replicates.
SpeB protease activity assay. SpeB protease activity was measured using fluorescein isothiocyanate (FITC)-labeled casein (G-Biosciences, St. Louis, MO) as a protease substrate. Supernatants of GAS strains grown overnight (ϳ16 h) in THY broth were obtained by centrifugation (4,000 ϫ g, 4°C, 10 min), diluted in the same volume of freshly prepared activation buffer (1 mM EDTA, 20 mM dithiothreitol [DTT], 0.1 M sodium acetate, pH 5.0), and incubated for 30 min at 40°C. Subsequent addition of the same volume of activation buffer containing 0.02 g/l FITC-labeled casein was followed by further incubation for 30 min at 40°C. Proteolytic activity of SpeB against FITC-labeled casein was measured by fluorescence (485-nmwavelength excitation, 535-nm-wavelength emission). Activity data shown represent means and standard errors of the means derived from triplicate determinations of samples and are representative of results from at least three independent experiments.
Complementation of cls and liaF deletion mutants. The cls and liaF genes, including their promoters, were PCR amplified (Table S3) from MGAS10870, digested with BamHI and with PstI, and ligated into the expression plasmid pLZ12Km2, resulting in pLZ12Km2::CLS or pLZ12Km2::LiaF (Table S2). For complementation studies, the mutant strains were transformed with plasmid pLZ12Km2::CLS or pLZ12Km2::LiaF, which carried an open reading frame (ORF) of the cls gene or liaF gene, respectively, under the control of its native promoter. The positive mutant colonies were selected on THY plates containing 150 g ml Ϫ1 kanamycin. The identity of the clones was confirmed by PCR and sequencing.
Detection of LiaR phosphorylation levels in vivo. LiaR phosphorylation levels were determined using Phos-Tag. GAS lysates were prepared using a FastPrep-24 5G homogenizer (MP Biomedicals) and separated on 12.5% SuperSeq Phos-tag gels (Wako, USA). Un/phosphorylated LiaR species were detected using a polyclonal anti-LiaR antibody (Covance) and a ChemiDoc MP imaging system (Bio-Rad). All phosphorylation experiments were performed at least twice on independent samples.
Statistics. All statistical analyses were performed with Prism 8 software (GraphPad Software, San Diego, CA). Differences between mean values (continuous variables) were determined using a two-tailed Student's t test. A P value of Ͻ0.05 was considered significant for all tests.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.

ACKNOWLEDGMENTS
We thank Samuel A. Shelburne III and Jovanka Voyich for comments and suggestions to improve the manuscript.