Engineered Bacillus subtilis Systems for the Characterization and Detection of Peptide-Based Quorum Sensing in Gram-Positive Bacteria Authors: Emma L. Eisenbraun, Natalie L. Beebe, Alexandra E. Nelson, Brendan N. Prosser, Troy D. Vulpis, Amanda E. Appel, Helen E. Blackwell Published: 2026 ABSTRACT Many questions remain about the biochemical mechanisms of quorum sensing (QS), particularly in Gram-positive bacterial pathogens that pose serious threats to human health such as Staphylococcus aureus. Methods to isolate and study the molecular components of these signaling systems in vitro are not straightforward. To date, most mechanistic investigations into QS systems have been accomplished in cells and, to a lesser extent, using chemical inhibitors. Herein, we report the development and characterization of a set of engineered Bacillus subtilis strains that can be used to study the accessory gene regulator (agr) QS circuits of a variety of Gram-positive bacteria. We cloned the S. aureus agr QS machinery into B. subtilis as proof-of-concept to generate “sender” cells capable of autoinducing peptide (AIP) signal biosynthesis, “receiver” cells capable of signal transduction, and “full” cells containing the full agr QS system. We verified that the B. subtilis sender cells could produce the native S. aureus AIP and that the receiver and full cells responded to both the AIP signal and a panel of known chemical agr modulators. The approach was readily transferable to study the agr QS systems of Staphylococcus epidermidis and Listeria monocytogenes, revealing interesting differences that could indicate underlying divergences in native QS mechanisms. We also demonstrate the ability of these B. subtilis strains to function as biosensors to detect the native AIPs of bacteria and produce a targeted antibiotic in response. These engineered systems should find utility for the study of QS in a range of fundamental and applied contexts. Introduction Quorum sensing (QS) is a cell-to-cell communication pathway that allows bacteria to regulate gene expression according to cell density. These genes primarily control group behaviors, such as virulence factor production, biofilm formation, motility, and sporulation. , Many of these behaviors play prominent roles in both acute and chronic bacterial infections. Because of this regulatory connection, targeting QS pathways has attracted considerable attention as a means of controlling and reducing bacterial virulence without affecting growth, offering a potential antivirulence strategy that contrasts with traditional antibiotics. Many common Gram-positive bacterial pathogens that pose threats to human and animal health regulate virulence at least in part using QS. One such bacterium is Staphylococcus aureus, the leading cause of endocarditis, bacteremia, osteomyelitis, and skin and soft tissue infections in humans. The prevalence of methicillin-resistant S. aureus (MRSA) in these infections poses an increasingly serious threat and medical burden. Another prominent Gram-positive bacterium that uses QS is Staphylococcus epidermidis, an opportunistic pathogen with a propensity to form biofilm infections on abiotic surfaces such as indwelling medical devices. These biofilms promote antibiotic resistance and elicit host immune responses, resulting in S. epidermidis emerging as a common cause of hospital-acquired infections. Listeria monocytogenes represents a third well-known and QS-active Gram-positive pathogen that is responsible for the foodborne illness listeriosis and accounts for 19% of all foodborne illness-related deaths in the United States. Further characterization of the QS systems in these three species could provide new pathways toward strategies to mitigate human infection and associated suffering. One of the best-characterized QS systems in Gram-positive bacteria is the accessory gene regulator (agr) system, which is utilized by the three pathogens introduced above to initiate a broad range of virulent behaviors. The agr QS system is best understood in S. aureus (Figure A) and is comprised primarily of four proteinaceous components: two proteins involved in QS signal biosynthesis (AgrB and AgrD) and two proteins involved in signal transduction (AgrC and AgrA). Briefly, the propeptide AgrD is processed by the transmembrane protease AgrB, which cleaves the AgrD C-terminal domain and facilitates macrocycle formation, creating the pro-autoinducing peptide (AIP). The N-terminal amphipathic leader sequence of the pro-AIP is then cleaved by a membrane protease (identified as MroQ in some subspecies [i.e., specificity groups] of S. aureus), to yield the resulting AIP signal. The AIP, which consists of a five-membered thiolactone macrocycle and a short N-terminal tail, is exported out of the cell by an unknown mechanism. Once cellular density has sufficiently increased, a threshold concentration of AIP is achieved in the local environment allowing for productive AIP binding to the extracellular sensor domain of its target receptor, the transmembrane histidine kinase AgrC. AIP:AgrC binding causes autophosphorylation of AgrC, which subsequently phosphorylates its corresponding cytoplasmic response regulator, AgrA. , Upon phosphorylation, AgrA dimerizes and binds at the divergent P2 and P3 promoters. , Binding at the P2 promoter upregulates the production of the agr machinery, resulting in a positive feedback cycle (or “autoinduction”) that is a hallmark of QS systems. In turn, AgrA binding at the P3 promoter activates the production of the effector molecule RNAIII, which is responsible for the direct and indirect regulation of an arsenal of virulence factors. For S. aureus and other closely related species, several divergent agr specificity groups exist (in S. aureus, groups I–IV), each with unique AgrB, AgrD, and AgrC components and distinct AIP signals, but with highly similar AgrA response regulators. These specificity groups exhibit varying degrees of agr interference (i.e., inhibition of nonself groups) via their AIP signals. , Schematic of the agr QS system and three types of Bacillus subtilis systems developed in this study. (A) The agr system in S. aureus. AIP is biosynthesized through the processing of the propeptide AgrD via the membrane protease AgrB and (in certain groups) MroQ. AIPs bind to the AgrC receptor, which activates the response regulator AgrA, upregulating the expression of the agr operon and the RNAIII effector molecule, responsible for regulating a variety of virulence factors. (B) Top: The B. subtilis sender cells contain xylose-inducible AgrBD machinery, capable of AIP biosynthesis. Middle; The B. subtilis receiver cells contain xylose-inducible AgrCA machinery, capable of receiving QS signals and reading out activity using GFP. Bottom: The B. subtilis full cells contain both biosynthesis and signal transduction machinery, capable of AIP generation and QS signal detection with fluorescence readout upon induction with xylose. Since the discovery of the native AIP signals and their interference, there has been considerable interest in the development of chemical ligands capable of inhibiting agr QS systems and thereby associated bacterial virulence. − A significant number of peptides, synthetic small molecules, and natural products have been discovered that are reported to inhibit various components of the S. aureus agr QS system. In S. epidermidis, a small number of non-native peptide agr activators and inhibitors have been developed, − and in L. monocytogenes, an even more limited set of peptide and small molecule agr inhibitors have been explored. − Representative structures are shown in Figure . While progress continues in the design of potent and efficacious QS modulators (QSMs), the mechanisms of the majority of these compounds are underexplored, including identifying their specific agr targets. Nevertheless, QSMs, of both natural and synthetic origins, represent valuable tools to investigate agr as a target for infection control in a range of cell-based and in vivo infection models ,,,, and can complement genetic approaches in such investigations. Panel of native and non-native compounds used in this study to characterize the B. subtilis constructs. Compounds were previously identified to modulate agr in (A) S. aureus, (B) S. epidermidis, and/or (C) L. monocytogenes. Letters in circles indicate amino acids with standard abbreviations. See text for citations to each compound. Heterologous or engineered QS systems, in which the signaling machinery can be studied outside of the native host cell environment, can provide multiple advantages for mechanistic studies of QSMs, including the separation of the QS system from pathways that intersect with it in the host background, the enhanced genetic tractability of many heterologous organisms, and the manipulation of (typically) a BSL1 organism. Much of the past work on heterologous QS systems has focused extensively on LuxI/LuxR-type QS pathways in Gram-negative bacteria. Far less has been reported on the development of heterologous Gram-positive QS systems, due in part to the lack of mechanistic information about many of these QS pathways and challenges associated with the genetic manipulation of Gram-positive bacteria. In 2013, the Collins laboratory reported one of the first synthetic QS systems based on that of a Gram-positive bacterium. In a series of studies, the authors were able to engineer Escherichia coli and Priestia megaterium that expressed components of the S. aureus agr system. , The Collins team created “receiver cells” that expressed AgrC and AgrA, along with “sender cells” that expressed AgrB and AgrD from group-I S. aureus. They also developed a “full cell” system where the full agr operon was placed in P. megaterium and showed that monocultures were capable of significant activation. , The goal of the Collins lab was not to explore the mechanisms of agr QS with these engineered strains, but rather to demonstrate proof-of-concept that the engineered strains could be used as tools for synthetic biology. Because of this focus, the authors only tested the ability of the sender or full cell supernatants to activate the corresponding receiver cells and did not characterize or quantitate the amount of AIP product generated. Since these initial reports, several other groups have introduced the S. aureus agr QS machinery into heterologous strains, including B. subtilis, , S. epidermidis, Lactobacillus reuteri, Lactobacillus plantarum, Clostridium sporogenes, and E. coli, for the purposes of creating biosensors, exploring delivery mechanisms of select agents, or modulating the induction of protein expression. These past reports demonstrate important advancements, including the fine-tuning of autoinduction, promoter strength, fluorescent readout, and temporal dynamics of the systems. , One notable advance involved expressing S. aureus agr components (in L. plantarum and S. epidermidis) to drive production of the antibacterial enzyme lysostaphin in response to AIP-I detection, demonstrating targeted killing of S. aureus. , Overall, these past studies were largely similar to those of the Collins group: most focused solely on S. aureus agr group-I and not on mechanistic questions about agrBDCA function. None of these studies examined QSMs outside of native signaling molecules, and apart from the work in L. reuteri, L. plantarum, and C. sporogenes, these studies neither confirmed the structure nor quantified the amount of AIP product generated in the heterologous systems. ,,, Herein, we report the development and characterization of a new suite of synthetic agr QS systems in B. subtilis. We first introduced the S. aureus agr machinery to this platform to establish our methods and generate sender, receiver, and full cells (Figure B). AIP production by the sender cells was verified and quantified using supernatant assays and mass spectrometry (MS) analysis. The receiver and full cells were used to evaluate a library of known synthetic S. aureus QSMs, which demonstrated activity profiles in these heterologous systems that helped to illuminate their mechanisms of agr modulation in wild-type S. aureus. The engineered platform was readily adapted to produce agr components from both S. epidermidis and L. monocytogenes, and similar characterizations of AIP production and response to QSMs were conducted. To our knowledge, this work represents the first report of the study of S. epidermidis and L. monocytogenes agr QS components outside of their native host. Lastly, we developed a biosensor system in B. subtilis using the S. aureus agr machinery that can detect three of the four native S. aureus AIPs and trigger the production of the antibacterial enzyme lysostaphin in response. This B. subtilis platform system offers a straightforward approach to explore both agr QS mechanisms and the activities of QSMs in Gram-positive species, and to detect and respond to a broad set of AIP signals, and thereby, pathogenic bacteria. Experimental Section Materials All standard chemical and biological reagents and solvents were purchased from commercial suppliers and used according to the instructions. Water (18 MΩ) was purified using an Arium Pro ultrapure water system (Sartorius). All solid-phase extraction solvents (water and acetonitrile (CH3CN)) were HPLC grade (Sigma-Aldrich). All native AIPs, analogues, and small molecules previously reported by our laboratory were acquired from in-house stocks and synthesized according toreported methods. ,,, S. aureus AIP-III + Ala (used as an internal standard in MS studies) was synthesized according to our previous methods; HPLC and MS characterization data are in the SI. Savirin was purchased from AK Scientific. Bacterial Strains and Growth Methods All bacterial strains used in this study are listed in Table . B. subtilis containing either S. aureus or L. monocytogenes agr circuitry and L. monocytogenes were grown in Luria–Bertani medium (LB, Research Products International). B. subtilis containing S. epidermidis agr circuitry and S. epidermidis were grown in Tryptic Soy Broth (TSB NutriSelect Basic, Sigma-Aldrich). S. aureus was grown in Brain Heart Infusion (BHI, Teknova). B. subtilis strains carrying a resistance marker were selected using erythromycin (1 μg/mL) combined with lincomycin (25 μg/mL), and S. aureus reporter strains (AH1677 and AH430) were selected using chloramphenicol (10 μg/mL). All bacterial cultures were grown at 37 °C with shaking at 200 rpm unless otherwise noted. A 10% w/v xylose (Sigma-Aldrich) stock was made in either LB or TSB medium and diluted to induce gene expression. 1Bacterial Strains and Plasmids Used in This Studynamedescriptionref or sourceStrains S. aureus USA300 LACClinical isolate, agr group-IOtto S. aureus USA300 LAC Δagr USA300 LAC, agrBDCA knockoutLee et al. S. aureus RN6390BWild-type S. aureus lab strain, agr group-IJi et al. S. aureus RN6925Wild-type S. aureus clinical isolate, agr group-IIJi et al. S. aureus MW2Wild-type S. aureus USA400, agr group-IIIKing et al. S. aureus RN4850Wild-type S. aureus lab strain, agr group-IVLyon et al. S. aureus AH1677USA300 LAC/pDB59 (Camr)Yarwood et al. and Kirchdoerfer et al. S. aureus AH430SA502A/pDB59 (Camr)Yarwood et al., Malone et al. and Kirchdoerfer et al. S. epidermidis RP62AClinical isolate, agr group-IGill et al. S. epidermidis 1457Wild-type S. epidermidis, agr group-IIMack et al. S. epidermidis 8247Wild-type S. epidermidis, agr group-IIIOlson et al. L. monocytogenes EGD-eWild-type L. monocytogenes lab strain (serovar 1/2a)Glaser et al. E. coli DH5αCloning strainNEB B. subtilis 168trpC2 lab strainBGSCb PlasmidspBS0EXylRPxylA(V2) Empty vector, shuttle vector for cloning in E. coli (AmpR), xylose-inducible expression in B. subtilis (Mlsr)Popp et al. BGSC pXyl-Sa-agrBDDerivative of pBS0EXylRPxylA(V2) with S. aureus AgrBD added for xylose-inducible expression (Mlsr)This reportpXyl-Sa-agrCA-GFPDerivative of pBS0EXylRPxylA(V2) with S. aureus AgrCA added for xylose-inducible expression and GFP added under P3 expression (Mlsr)This reportpXyl-Sa-agrCA-LysTDerivative of pBS0EXylRPxylA(V2) with S. aureus AgrCA added for xylose-inducible expression and lysostaphin added under P3 expression (Mlsr)This reportpXyl-Sa-agrBDCA-GFPDerivative of pBS0EXylRPxylA(V2) with S. aureus AgrBDCA added for xylose-inducible expression and GFP added under P3 expression (Mlsr)This reportpXyl-Se-agrBDDerivative of pBS0EXylRPxylA(V2) with S. epidermidis AgrBD added for xylose-inducible expression (Mlsr)This reportpXyl-Se-agrCA-GFPDerivative of pBS0EXylRPxylA(V2) with S. epidermidis AgrCA added for xylose-inducible expression and GFP added under P3 expression (Mlsr)This reportpXyl-Se-agrBDCA-GFPDerivative of pBS0EXylRPxylA(V2) with S. epidermidis AgrBDCA added for xylose-inducible expression and GFP added under P3 expression (Mlsr)This reportpSa-control-GFPDerivative of pXyl-Sa-agrBDCA-GFP with the AgrBDCA gene removed (Mlsr)This reportpSe-control-GFPDerivative of pXyl-Se-agrBDCA-GFP with the AgrBDCA gene removed (Mlsr)This reportpLm-control-GFPDerivative of pXyl-Lm-agrBDCA-GFP with the AgrBDCA gene removed (Mlsr)This reportpXyl-Lm-agrBDDerivative of pBS0EXylRPxylA(V2) with L. monocytogenes AgrBD added for xylose-inducible expression (Mlsr)This reportpXyl-Lm-agrCA-GFPDerivative of pBS0EXylRPxylA(V2) with L. monocytogenes AgrCA added for xylose-inducible expression and GFP added under P3 expression (Mlsr)This reportpXyl-Lm-agrBDCA-GFPDerivative of pBS0EXylRPxylA(V2) with L. monocytogenes AgrBDCA added for xylose-inducible expression and GFP added under P3 expression (Mlsr)This reportaNew England Biolabs.bBacillus Genetic Stock Center. Bacterial Strains and Plasmids Used in This Study New England Biolabs. Bacillus Genetic Stock Center. Plasmid Construction All plasmids used and created in this report are listed in Table . Primers used in plasmid construction were ordered from Integrated DNA Technologies and are listed in Table S1. Vectors in this study were generated using standard cloning techniques, including a combination of Gibson Assemblies using Phusion polymerase chain reaction (PCR) and Q5 site-directed mutagenesis methods. Full details of plasmid construction, including primer and vector sequences, are included in the Supporting Information (Figure S1, Tables S1, and S2). Briefly, agr machinery was cloned from S. aureus, S. epidermidis, and L. monocytogenes genomic DNA (extracted from RN6390B, RP62A, and EGD-e, respectively). These genes were added under control of a xylose-inducible promoter to the vector pBS0EXylRPxylA(V2), along with a B. subtilis ribosomal binding site (RBS). In the receiver cells and the full cells, a B. subtilis-optimized GFP gene was added under the control of the S. aureus, S. epidermidis, or L. monocytogenes P3 promoter. For the biosensor system, lysostaphin fused to an AmyE secretion tag replaced GFP downstream of the S. aureus P3 promoter. Constructs that isolate components of the agr operon or remove the agr operon were created for all three agr systems to serve as negative controls. All vectors were characterized using Plasmidsaurus sequencing (see SI for full vector sequences). Plasmids were transformed into B. subtilis 168 cells using methods published by Bennallack et al., generating the final sender cells (containing pXyl-agrBD), receiver cells (containing pXyl-agrCA-GFP), full cells (containing pXyl-agrBDCA-GFP), and biosensor cells (containing pXyl-Sa-agrCA-LysT), along with the controls (containing pXyl-control-GFP, pXyl-agrA-GFP, and pXyl-agrC-GFP). Growth curves for all engineered strains can be found in the SI (Figure S2). B. subtilis Supernatant Assays To isolate supernatants, overnight cultures of B. subtilis sender cells or wild-type (WT) B. subtilis 168 were diluted 1:100 in LB or TSB medium supplemented with 1% xylose and grown at 37 °C for 8 h at 200 rpm. Cells were then pelleted by centrifugation at 4 °C for 15 min at 4000 rpm, and supernatant was filtered using Steriflip filter units (Millipore, 0.22 μm). Supernatant was frozen with liquid nitrogen and lyophilized for 36 h. Lyophilized supernatants were resuspended in sterile water at either 1/10 or 1/25 of their original volume. To test the 10-fold or 25-fold concentrated B. subtilis sender cell supernatants, they were serially diluted in WT B. subtilis 168 supernatant. Aliquots (20 μL) of the diluted supernatant were added to each of the wells of a black 96-well polystyrene microtiter plate (Costar). The diluted supernatant samples from B. subtilis sender cells based on S. aureus and L. monocytogenes agr were tested in technical and biological triplicate, and supernatant samples isolated from B. subtilis sender cells based on S. epidermidis agr were tested in technical duplicate and biological triplicate. A 20 μL-aliquot of WT B. subtilis supernatant was included as a vehicle control. Overnight cultures of S. aureus fluorescent reporter strains (AH1677 and AH430) were grown for 20 h and diluted 50-fold with fresh BHI medium. Aliquots (180 μL) of the diluted reporter cultures were added to each of the wells of the microtiter plate containing supernatant. Plates were then incubated at 37 °C for 24 h at 750 rpm on a Stuart SI505 microtiter plate shaker incubator, and the fluorescence (Excitation 500 nm/Emission 540 nm) and OD600 of each well were measured using a Biotek Synergy 2 plate reader running Gen5 3.12 software. Data were normalized to the vehicle (100%) and media (0%) controls and then analyzed using GraphPad Prism software (v. 10.3.1) to generate sigmoidal curve fits ([inhibitor] vs response, four-parameters). B. subtilis Fluorescent Reporter Assays QSM agonism and antagonism assays in B. subtilis receiver and full cells were performed using previously established protocols with slight changes. Compound stock solutions in DMSO (1–10 mM) were serially diluted 3-fold with DMSO or left undiluted for point concentration experiments. Aliquots (2 μL) of the compound solutions were added to each of the wells of a black 96-well polystyrene microtiter plate (Costar). Each compound was tested in technical and biological triplicate unless otherwise noted, and 2 μL of DMSO was included as a vehicle control. For the B. subtilis receiver and full cells containing the S. aureus or L. monocytogenes agr systems, overnight cultures were grown for 20 h and diluted 50-fold with fresh LB medium, supplemented with 1% w/v xylose, and 198 μL aliquots of the diluted culture were added to each of the wells of the microtiter plate containing compound. For the B. subtilis receiver and full cells containing the S. epidermidis agr systems, overnight cultures were grown for 20 h and diluted 50-fold with fresh TSB medium, supplemented with 0.01% w/v xylose, and 198 μL aliquots of the diluted culture were added to each of the wells of the microtiter plate containing compound. Plates were then incubated at 37 °C for 24 h at 750 rpm on a Stuart SI505 microtiter plate shaker incubator, and the fluorescence (Excitation 500 nm/Emission 540 nm) and OD600 of each well were measured using a Biotek Synergy 2 plate reader running Gen5 3.12 software. Data were normalized to growth and media controls, and some full cell data was normalized to the vehicle (100%) and media (0%) controls. Data was analyzed using GraphPad Prism software (v. 10.3.1) to generate sigmoidal curve fits ([compound] vs response, four-parameters). Significance testing was performed using one-way or two-way ANOVA with Šídák’s multiple comparisons test or an unpaired t test. Differences with **, ***, and **** were deemed significant with p adj < 0.01, p adj < 0.001, and p adj < 0.0001, respectively. Control B. subtilis cells that contain the xylose-inducible vector with the agr gene removed, but with the P3-GFP readout component present (pSa-control-GFP, pSe-control-GFP, pLm-control-GFP), were tested and did not show significant GFP production (Figure S3). Mass Spectrometric Analysis of Supernatants for AIP Detection B. subtilis supernatant samples were isolated as described above. Following centrifugation and filtration, samples were acidified with trifluoroacetic acid (TFA) to 0.1% and spiked with an internal standard (S. aureus AIP-III + N-terminal alanine; sequence AIN-(CDFLL)) to 5 μM before solid-phase extraction (SPE) of native AIPs. SPE of AIP was conducted as described previously with minor modifications. Briefly, C18 SPE columns (SiliCycle SiliaPrep SPE CartridgesC18 (17%) poly) were activated with 3 × 3 mL of a 10%/90%/0.1% mixture of H2O/CH3CN/TFA. Columns were then equilibrated with 2 × 3 mL of 0.1% (v/v) TFA in water. Supernatant samples were loaded onto the column and washed with 5 mL of a 91%/9%/0.1% mixture of H2O/CH3CN/TFA. Peptides were eluted with 5 mL of a 10%/90%/0.1% mixture of H2O/CH3CN/TFA and lyophilized overnight prior to LC-MS analysis. All mass spectrometry (MS) analyses were conducted using a Q-Exactive Focus ESI-Q-Orbitrap Mass Spectrometer (Thermo Scientific) equipped with an UltiMate 3000 HPLC (Thermo Scientific) and diode array detector. Dried SPE eluent was resuspended in 50% HPLC grade CH3CN in water such that the final eluted sample was concentrated 10× (relative to initial supernatant sample) and filtered using a 0.45 μM syringe filter (PerkinElmer Red PTFE (Hydrophilic) Syringe Filter). Concentrated samples (5 μL) were injected onto an Agilent PLRP-S 300A (50 mm × 2.1 mm) column using a 10–60% gradient of CH3CN in water with 0.1% formic acid (FA) over 8 min. The masses of native AIPs and internal standard were added to the inclusion list as precursor ions for parallel reaction monitoring (PRM). Extracted ion chromatograms (MS1) of S. aureus AIP-I ([M + H]+ = 961.3794), L. monocytogenes 6-mer ([M + H]+ = 699.3000), and the internal standard ([M + H]+ = 890.4440) were integrated. Peaks were integrated within a 5-ppm window around the calculated mass and normalized to the internal standard peak area. S. aureus AIP-I and L. monocytogenes 6-mer were quantified using a standard curve generated from authentic samples of synthetic AIP (Figure S4). Standard curve data were fitted to a simple linear regression model. All peptide masses and identities were confirmed by matching parent m/z, MS2 (assisted by PRM), and retention time to synthetic standards. S. aureus AIP-I produced by B. subtilis showed relatively poor fragmentation under these conditions; only four fragment ions were used to confirm its identity. B. subtilis Biosensor Zone of Inhibition Assays Overnight cultures of WT B. subtilis 168 and the lysostaphin B. subtilis construct (SaRC-LysT) were grown in LB medium and supplemented with no antibiotic or erythromycin (1 μg/mL) + lincomycin (25 μg/mL), respectively. Overnight cultures of S. aureus (LAC, LAC Δagr, RN6925, MW2, and RN4850) or S. epidermidis (RP62A, 1457, and 8247) were grown in TSB medium. The overnight cultures were then diluted 1:500, and a 200-μL aliquot of culture was spread evenly with a sterile swab on tryptic soy agar plates (100 × 15 mm) supplemented with 0.1% xylose and allowed to dry. The cultures of B. subtilis 168 and SaRC-LysT were pelleted via centrifugation (13,200 rpm for 1 min) and washed twice in 1 mL of PBS to remove antibiotics from the supernatant. Aliquots (5 μL) of washed B. subtilis 168 and SaRC-LysT cells were spotted onto the lawn and allowed to dry. Plates were incubated statically at 37 °C for 24 h, after which they were imaged (using an iPhone 14 camera), and zones were measured using ImageJ. Significance testing was performed analogously to that described for the B. subtilis fluorescent reporter assays. B. subtilis Biosensor MIC Supernatant Assays Overnight cultures of S. aureus LAC were grown in TSB medium at 37 °C for 20 h at 200 rpm. B. subtilis constructs (SaRC-LysT and controls) were grown in LB medium supplemented with various amounts of xylose in either the presence or absence of 1 μM S. aureus AIP-I at 37 °C for 20 h at 200 rpm. An additional culture of WT B. subtilis 168 was grown in LB medium under the same conditions. All B. subtilis cultures were pelleted via centrifugation for 2 min at 13,200 rpm. Supernatants were collected and filtered through a 0.22 μm syringe filter. S. aureus LAC overnight cultures were then diluted 1:1000 in fresh TSB medium, and 50 μL aliquots were transferred to the wells of a clear 96-well microtiter plate (Costar) along with 50 μL of B. subtilis supernatant. Plates were incubated statically at 37 °C for 24 h before the OD600 of each well was measured on a Biotek Synergy 2 plate reader running Gen5 3.12 software. Supernatants were tested in five biological replicates in technical duplicate. Measurements were background corrected and normalized to a 100 μL S. aureus LAC control (grown in the absence of B. subtilis supernatant). Data were analyzed and plotted using GraphPad Prism software (v. 10.4.2). Significance testing was performed analogously to that described for the B. subtilis fluorescent reporter assays. Results and Discussion Development of a S. aureus agr “Sender” System in B. subtilis We began by cloning S. aureus agrBD into B. subtilis to yield “sender" cells (referred to hereafter as Sa sender cells for brevity) capable of biosynthesizing AIP. To start, the S. aureus agrBD gene was amplified using PCR from extracted genomic S. aureus DNA (group-I, strain RN6390B), added to the backbone vector (pBS0ExylRPxylA(V2)), and placed under control of a xylose-inducible promoter, using Gibson Assembly methods. A B. subtilis ribosomal binding site was added using primers directly preceding the agr genes, generating the final pXyl-Sa-agrBD vector. Sequence-confirmed vectors were then transformed into B. subtilis 168 cells (see Experimental section for full details of methods and plasmid construction). Work from the Collins group showed that expression of the agrBD genes in both E. coli and P. megaterium yielded supernatants (assumed to contain native AIP-I) capable of activating “receiver” cells containing the agrCA genes, suggesting that the final AIP N-terminal processing steps (performed by MroQ in S. aureus group-I) are likely performed by either an analogous protease or one(s) with similar specificity in these heterologous hosts. , At the outset, we could not discount the possibility that AIP products with various tail lengths would be generated, and that such signals could cause a response, as S. aureus AgrC-I has been shown to be activated by AIP-I analogs with varied N-terminal tails. To test if the Sa sender cells were producing AIP-I, supernatant from sender cells induced with xylose and grown for 6–8 h was isolated, concentrated, serially diluted, and subsequently added to cultures of a S. aureus group-II agr reporter strain, AH430, which reads out agr activity as fluorescence (i.e., production of YFP is under control of the agr-regulated P3 promoter). AIP-I is known to strongly inhibit group-II agr. Thus, if S. aureus AIP-I was being produced at sufficient levels by the Sa sender cells, the supernatant should inhibit agr in this reporter strain. As anticipated, we observed a dose-dependent inhibition of group-II agr upon the addition of the Sa sender cell supernatant (Figure A), reaching 96% inhibition at a supernatant concentration of 250% (unconcentrated supernatant set to 100%). Characterization of the engineered B. subtilis strains containing the S. aureus agr QS system. (A) Dose–response of Sa sender cell supernatant in a S. aureus group-II agr fluorescent reporter strain (AH430). Data normalized to media (0% activity) and WT B. subtilis 168 supernatant (100% activity) controls. (B) Mirror plot of MS2 spectra comparing Sa sender cell supernatant (blue, RT = 6.30 min) compared to a synthetic S. aureus AIP-I standard (black, RT = 6.30 min). Labeled mass identities represent peaks identified in both samples. (C) Fluorescence readout of the Sa receiver and Sa full cells with and without xylose induction, and upon the addition of 10 μM S. aureus AIP-I. (D) Activity data for a panel of QS inhibitors (and native agonist AIP-I) in Sa receiver cells. All concentrations are 10 μM. Dotted line represents vehicle fluorescence for ease of comparison. Fluorescence normalized to growth and media controls. Activity not reported as percent activation in panels (C) or (D) due to the lack of basal AIP production in the receiver cells that is usually standardized as “100% activity.” (E) Activity data for a panel of QS inhibitors (and native agonist AIP-I) in Sa full cells. All concentrations are 10 μM. Dotted line represents vehicle fluorescence for ease of comparison. Fluorescence normalized to media (0% activity) and DMSO (100% activity) controls. Error bars in each plot indicate the standard error of the mean (SEM) of at least three biological replicates, each tested in technical triplicate (n = 3). Significance values represent comparisons to DMSO control. See Experimental section for full details of methods. For further verification of native AIP-I production, we analyzed Sa sender cell supernatants using a liquid chromatography-tandem mass spectrometry (LC-MS/MS) method and compared the results to a S. aureus AIP-I synthetic standard. We observed that the major peak in the Sa sender cell supernatant matched the retention time, molecular ion mass to charge ratio (m/z), and MS/MS fragmentation patterns of the AIP-I standard (Figure B, and S5). A standard curve of synthetic S. aureus AIP-I (Figure S4) was used to quantify the amount of AIP in sender cell supernatant (at 8 h), and the sender cells were found to be producing approximately 11 nM AIP-I. While the major peak in the Sa sender cell supernatant corresponded to AIP-I, an additional peak was also observed corresponding to the m/z of the AIP plus an extra alanine residue at the N-terminus (sequence AYST-(CDFIM), [M + H]+ = 1032.4165), indicating a mis-cleaved product. [It is possible that the activity of this minor mis-cleavage product is contributing to the overall activity profile of our Sa constructs; however, similar N-terminal extensions on AIP-I as observed in this byproduct have been shown to activate AgrC-I to similar levels as the native AIP. Therefore, we do not anticipate the production of this minor byproduct to confound interpretation of QSM activity in this Sa system.] Masses of AIP-I and its mis-cleaved byproduct were only observed in cells containing the agrBD construct and not in the B. subtilis 168 background control, suggesting these masses correspond to peptide products of AgrD processing and not endogenous B. subtilis metabolites. For comparison of sender cell vs native AIP production, supernatants of S. aureus USA300 LAC (group-I) cells grown for 6–8 h were similarly analyzed by LC-MS/MS. The S. aureus cells were found to produce significantly more AIP-I (347 nM) than the Sa sender cells (∼30-fold). Additionally, no major mis-cleavage products were observed in the S. aureus supernatant. This latter result suggests that while there is a protease in the B. subtilis sender cells capable of performing the N-terminal cleavage of the pro-AIP like MroQ in S. aureus, it may not operate with the same precision. We note that this cleavage step in the sender cells could be performed by the putative MroQ homologue in B. subtilis, YdiL, a different protease, or multiple proteases. Development of a S. aureus agr “Receiver” System in B. subtilis We next generated S. aureus “receiver" cells (Sa receiver cells for brevity) in B. subtilis, using similar cloning methods to those described above yet containing the S. aureus agrCA signal transduction system genes instead of the agrBD genes. Within the same vector, a B. subtilis-optimized GFP gene was added under control of the S. aureus P3 promoter using a Gibson Assembly to provide a means for measuring AgrCA activity (see Experimental section). We first examined fluorescence in the Sa receiver cells in the presence and absence of xylose, and with the native signal AIP-I (Figure C). As expected, minimal fluorescence was observed in the absence of xylose. Upon the addition of xylose, a significant increase (16-fold) in fluorescence was observed, primarily due to basal levels of AgrA binding to the P3 promoter (as determined by examination of a receiver cell control containing only agrA, Figure S6). Upon the addition of 10 μM AIP-I, fluorescence was further increased approximately an additional 2-fold, indicating activation of the AgrCA system (Figure C). A dose–response experiment was performed in the Sa receiver cells, showing a dose-dependent increase in fluorescence upon addition of increasing concentrations of AIP-I (Figure S7). Interestingly, the curve was flatter than typical sigmoidal dose–response curves in S. aureus agr reporter systems (with a Hill slope of 0.57), suggestive that AgrC receptor cooperativity could be different in the non-native background. The only reported analysis of S. aureus AgrC cooperativity was, to our knowledge, performed by the Muir lab in 2014, where they examined the response of AgrC-I isolated in nanoscale lipid discs (“nanodiscs”) to various ligands. In these prior in vitro experiments, AgrC-I was found to lack cooperativity upon activation by AIP-I. We note that the flatter dose–response curve observed in our B. subtilis receiver cells relative to S. aureus could occur for several reasons other than cooperativity differences, including the different expression levels of AgrC or AgrA, as overproduction of protein could result in a shift in the target to inhibitor ratio, and/or protein structure differences in the heterologous background (see SI for additional data (Figure S8) and discussion). We next tested a panel of known S. aureus group-I agr inhibitors in the Sa receiver cells to ascertain whether the compounds maintained their activity profiles in the non-native background, including the native peptide inhibitors AIP-II and AIP-III, the non-native peptide inhibitors AIP-III D4A (an inhibitor based on the native AIP-III scaffold, D4A for brevity), 12, and AAA (two derivatives of S. epidermidis native AIPs that are potent and efficacious pan-group inhibitors in both S. epidermidis and S. aureus), and the small molecule inhibitor savirin (see Figure for structures and Table S3 for compound activity in a WT S. aureus agr group-I reporter). These peptidic compounds are all believed to act on AgrC, while savirin has been shown to act on AgrA. We observed that the majority of these compounds decreased fluorescence in the Sa receiver cells, as expected (Figure D). However, the resulting fluorescent signals were below basal (DMSO vehicle) levels, suggesting that these inhibitors have inverse agonist activity profiles in this B. subtilis system. This inverse agonist profile for S. aureus AIP-II has been observed before by the Muir lab in their in vitro studies of AgrC-I. Interestingly, two of the inhibitors, the native AIP-III and D4A, increased fluorescence above vehicle levels, albeit less so than the native agonist AIP-I (AIP-III and D4A with 1.55-fold and 1.37-fold activation, respectively, vs AIP-I with 1.8-fold activation above basal DMSO activity). There is some precedence for this unexpected agonistic behavior for these two compounds, again in the context of in vitro experiments. Again, the Muir lab reported activation of AgrC-I in vitro upon the addition of the inhibitors AIP-III and AIP-I D5A (an inhibitor based on the AIP-I scaffold, but with a similar alanine substitution as in D4A). This 2015 result is congruent with our findings here in the Sa receiver system and, along with the observed inverse agonism profile for at least AIP-II, suggests that certain agr modulators are interacting with the AgrC-I receptor similarly in the non-native B. subtilis and in vitro nanodisc systems, yet differently relative to the native environment of the S. aureus agr reporters. We note that the native S. aureus membrane is more negatively charged and composed of different ratios of charged lipids and other membrane components relative to that of B. subtilis, , potentially contributing to the observed differences in compound activity. Development of a S. aureus agr “Full” System in B. subtilis Building on the results above, B. subtilis producing the “full” S. aureus agr system was developed and characterized (i.e., Sa full cells). These B. subtilis cells contained the entire S. aureus agrBDCA operon and the P3-controlled GFP readout system within the same vector (pXyl-Sa-agrBDCA-GFP), and were generated using similar cloning methods to the Sa sender and receiver cells. It is worth noting that these cells should mimic S. aureus agr reporter cells, with the ability to biosynthesize the native signal, sense QSMs, and read out their agr activity via fluorescence. The primary difference between the agr circuits in the B. subtilis full cells and the native S. aureus background is the use of xylose to induce agr expression in B. subtilis, as opposed to the cell density-induced positive QS feedback loop via the P2 promoter present in S. aureus. Thus, the QS dynamics and agr machinery production differ between the engineered B. subtilis strain and S. aureus as cell density changes. Control experiments were performed to test the activity of the Sa full cells with and without the addition of xylose and with S. aureus AIP-I. As observed for the receiver cells, minimal fluorescence was observed for uninduced cells (Figure C). Upon addition of xylose, a large increase in fluorescence was observed after 24 h, approximately double of that observed in the Sa receiver cells with added AIP-I. This result indicates that the full cells are both producing AIP and capable of receiving the native signal. Interestingly, a significant increase in fluorescence was not observed upon the addition of exogenous AIP-I (up to 10 μM) after 24 h, potentially indicating that all AgrC binding sites are already saturated with AIP-I produced endogenously by the full cells at that time point. The same panel of agr inhibitors that was tested in the Sa receiver cells was also tested in the Sa full cells (Figure E). Again, inhibition was generally observed across the board, with all AgrC inhibitors (aside from AIP-III) and the AgrA inhibitor savirin demonstrating statistically significant decreases in activity below the vehicle control (i.e., inverse agonist behavior). Select inhibitors were subjected to dose–response studies in the Sa full cells and found to block fluorescence in a concentration-dependent manner (Figure S7). No inhibition was observed for AIP-III, as observed in the Sa receiver cells, yet no activation was observed either (fluorescence similar to vehicle). At the 1% xylose condition tested, unexpected growth effects were observed for D4A in the Sa full cells that were not observed for any other compound. When the cells were induced at a lower xylose concentration (0.1%), these growth effects were no longer observed for D4A, but reduced inhibition levels were observed for all compounds (except for AIP-III (which was inactive) and savirin, Figure S9). Overall, the results so far demonstrated that these engineered B. subtilis strains can produce S. aureus AIP-I (via the sender and full strains) and respond to both native AIP-I and known agr inhibitors (via the receiver and full strains) and served to validate our design of heterologous S. aureus agr systems in B. subtilis. Application of the Engineered B. subtilis System to the S. epidermidis agr Circuit We were interested in whether this B. subtilis system could be applied to other agr circuits and turned our attention to S. epidermidis. The analogous combination of sender, receiver, and full cells were generated as for S. aureus, but now containing the group-I S. epidermidis agrBD, agrCA, and agrBDCA genes on the plasmids pXyl-Se-agrBD, pXyl-Se-agrCA-GFP, and pXyl-Se-agrBDCA-GFP (Table ), respectively, using the same cloning methods as described above. In addition, the S. epidermidis P3 promoter was used to control GFP production in the receiver and full cells, as opposed to the S. aureus P3 promoter. First, the cells containing the S. epidermidis agrBD genes were tested (i.e., the Se sender cells). In a similar manner to the Sa sender cells, the supernatant from the induced Se sender cells was isolated after 6–8 h, concentrated, and added to the S. aureus group-I agr fluorescence reporter, AH1677. , S. epidermidis AIP-I is known to inhibit S. aureus group-I agr (to 92.3%, IC50 = 74.4 nM), and thus if AIP-I was being successfully produced by the Se sender cells, its supernatant should inhibit agr in this reporter strain. While inhibition was observed, the level of inhibition was significantly less than that observed for the engineered Sa sender cells (Figure A), only reaching 34% inhibition at the highest concentration of supernatant tested. Furthermore, neither S. epidermidis AIP-I nor AIP miscleavage products could be detected in Se sender cell supernatant using analogous LC-MS/MS methods as described above for the Sa sender system. These results suggest that the Se sender cells may produce AIP-I (and/or miscleavage products) but at significantly lower levels than the AIP-I signal produced by the Sa sender cells. This result could be for a number of reasons. For example, AgrB/AgrD could be expressed at different levels in the Sa vs Se sender cells, or S. epidermidis AgrB/AgrD are less functional in the non-native membrane environment relative to S. aureus AgrB/AgrD. Alignment of the AgrB sequence from S. aureus and S. epidermidis used in the Bacillus systems shows 76% sequence similarity and 53% sequence identity, suggestive of some potential differences (Figure S10). Alignment of the two AgrD sequences shows 65% sequence similarity and 53% identity, with much of the variability existing in the N-terminal and AIP sequence regions (Figure S10). This variation in AgrD could cause the N-terminal leader sequence of S. epidermidis AgrD to associate differently with the B. subtilis membrane relative to S. aureus AgrD, leading to less cleavage and AIP production. Another reason for low AIP production could be because the (unknown) protease(s) responsible for the final N-terminal cleavage in AIP biosynthesis in the B. subtilis background are less facile at processing the S. epidermidis pro-AIP compared to the S. aureus pro-AIP. Characterization of the engineered B. subtilis strains containing the S. epidermidis agr QS system. (A) Dose–response of Se sender cell supernatant in a S. aureus group-I agr fluorescent reporter (AH1677). Data normalized to media (0% activity) and WT B. subtilis 168 supernatant (100% activity) controls. (B) Fluorescence readout of the Se receiver and full cells with and without xylose induction, and upon the addition of 1 μM S. epidermidis AIP-I. (C) Activity data for a panel of QS activators in Se receiver cells. All concentrations are 1 μM. For panels (C–F), fluorescence was normalized to growth and media controls. (D) Activity data for a panel of QS inhibitors (and native agonist AIP-I) in Se receiver cells. All concentrations are 1 μM. Activity not reported as percent activation in panels (B–D) due to the lack of basal AIP production in the receiver cells that is usually standardized as “100% activity”. (E) Activity data for a panel of QS inhibitors competed against 200 nM AIP-I in Se receiver cells. All inhibitor concentrations are 1 μM. Data normalized to media (0% activity) and 200 nM AIP-I (100% activity). (F) Activity data for a panel of QS inhibitors and the inactive compound Cmr3S6α in Se full cells. All concentrations are 1 μM. Dotted line represents vehicle fluorescence for ease of comparison. Error bars in each plot indicate the standard error of the mean (SEM) of at least three biological replicates, each tested in technical triplicate (n = 3). Significance values represent comparisons to DMSO control. See Experimental section for full details of methods. Next, we examined the receiver cells in B. subtilis containing the S. epidermidis AgrCA machinery and the fluorescent readout system under control of the P3 promoter (i.e., Se receiver cells). The receiver cells were first tested with and without xylose induction and with the native S. epidermidis AIP-I (Figure B). As observed for the Sa receiver system, minimal fluorescence was observed in the absence of xylose. In contrast to the Sa system, however, very little background fluorescence was observed upon xylose induction, potentially indicating that the basal level of AgrA binding to the P3 promoter is much lower in this Se receiver construct relative to the Sa system (as supported by examination of a receiver cell control containing only S. epidermidis agrA, Figure S6). Upon the addition of 1 μM S. epidermidis AIP-I to the Se receiver cells, a significant increase (over 30-fold) in fluorescence was observed, congruent with the AgrC and AgrA proteins being present and functional. The Se receiver cells next were tested against a panel of known peptidic S. epidermidis agr agonists, all of which are believed to act on AgrC. Unlike S. aureus, a significant number of non-native peptide-based agr agonists have been developed for S. epidermidis (e.g., AA, Cmr1, Cmr1S7A, 10, 8, Cmr1S7A-Ex10, and Cmr7F9­(3FPhe), structures in Figure ), , and we selected agonists for study capable of a wide range of agr activation levels (from 40 to 182%, as demonstrated in S. epidermidis agr reporters; see Table S3). Significant activation was observed in the Se receiver cells for all agonists tested at 1 μM (Figure C). Only minimal activity differences were observed between these compounds at this concentration; however, full dose–response curves of select compounds demonstrated clear discrimination between the compounds by the receiver cells (Figure S7C). These differences in activation levels followed trends similar to those observed in the S. epidermidis fluorescent reporter system (Table S3 and Figure S7C). , We also tested a panel of known peptidic S. epidermidis agr inhibitors (S. epidermidis AIP-II and AIP-III, AAA, 12, 5, and 6; structures in Figure ) in the Se receiver cells with high inhibitory efficacies in S. epidermidis agr reporters (87–97%) (Table S3). As expected, the majority of these compounds failed to show a statistically significant difference from the DMSO control and demonstrated significantly less activation than the native activator AIP-I, even at up to 1 μM. In addition, and similar to our results in the Sa receiver system, certain native AIPs or very close analogs, such as AIP-III and AAA (an inhibitor based on the S. epidermidis native AIP-I with threealanine substitutions) significantly activated above the DMSO control (Figure D). We sought to further probe whether these compounds were acting as inhibitors, as baseline activity in this receiver system could also simply indicate inactivity. We performed the same inhibition assay in the Se receiver cells but competed all the compounds against 200 nM AIP-I (a concentration well above that necessary for full activation) and, in contrast to the nonactivated Se receiver cell data, observed significant decreases in activity for all the compounds compared to 200 nM AIP-I alone (Figure E). Thus, significant activation of the system is needed to observe inhibitory activity of compounds, mimicking the native S. epidermidis systems in which the compounds were originally characterized. − Overall, these results indicate that the panel of S. epidermidis agr inhibitors are capable of significant antagonism in the Se receiver cells. We proceeded to generate and characterize a B. subtilis strain that contained the full S. epidermidis agrBDCA operon and the P3-GFP reporter plasmid (i.e., Se full cells). Control experiments were performed testing these cells with and without the addition of xylose and with S. epidermidis AIP-I (Figure B). As expected, and analogous to the Se receiver cells, minimal fluorescence was observed for uninduced cells. Upon addition of xylose, a significant increase (over 3-fold) in fluorescence was observed, indicative of a functional agr system. The fluorescence signal increased approximately 8-fold upon addition of 1 μM S. epidermidis AIP-I. This result was markedly different than that observed for the Sa full cells (Figure C), which exhibited no significant difference in fluorescence upon addition of S. aureus AIP-I, and is likely explained by the significantly lower AIP production observed in the Se sender cells compared to their Sa counterparts. These results could be congruent with the observed ability of S. epidermidis fluorescent agr reporters to activate over basal activation (set as 100%), a trend that is not observed in analogous S. aureus agr reporters. Lastly, the same set of S. epidermidis agr inhibitors tested in the Se receiver cells were tested in the Se full cells, along with peptide Cmr3S6α, which we have shown to have no measurable activity in S. epidermidis agr reporter strains and thus served as a valuable control. Every compound other than Cmr3S6α showed significant inhibition below vehicle levels of fluorescence in the full cells (Figure F). The trends in relative levels of inhibition in the Se full cells were identical with those observed in the Se receiver cells with added AIP-I (Figure E), with the strongest inhibition observed for S. epidermidis AIP-II and the least inhibition observed for AAA. Dose–response curves were collected on select inhibitors in the full cell system, and the trends in relative potencies matched those previously reported in S. epidermidis fluorescent agr reporters, although greater differences in efficacies were observed in the Se full cells (Figure S7). Overall, the results for the engineered B. subtilis Se cells illuminated some significant differences between how the S. aureus and S. epidermidis agr components behave in this heterologous background, particularly basal AIP production levels (low in S. epidermidis vs S. aureus) and background activity of the xylose-induced system prior to compound addition (low in S. epidermidis vs S. aureus). These observed differences between the two systems can be interpreted as potential intrinsic differences in the native agr signal networks in these closely related species; we return to this below. Nevertheless, the receiver and full cells for both S. aureus and S. epidermidis provide robust responses to both agr activators and inhibitors, further validating the design of these heterologous agr systems and underscoring their potential utility. Application of the Engineered B. subtilis System to the L. monocytogenes agr Circuit We were interested to study agr circuits beyond Staphylococcal species and expanded our approach to L. monocytogenes, a pathogen that poses a major threat to human health but has an underexplored agr QS system relative to S. aureus and S. epidermidis. B. subtilis sender, receiver, and full cells were constructed containing the AgrBD, AgrCA, and AgrBDCA machinery from L. monocytogenes (using plasmids pXyl-Lm-agrBD, pXyl-Lm-agrCA-GFP, and pXyl-Lm-agrBDCA-GFP, Table ), respectively, and analogous cloning techniques as described above. As with previous constructs, the L. monocytogenes P3 promoter was added in front of the GFP gene for fluorescence readout of agr activity. Similar to the two systems above, L. monocytogenes sender cells were created by adding the L. monocytogenes agrBD AIP biosynthesis genes to B. subtilis under control of a xylose-inducible promoter (Lm sender cells). Sender cell supernatant was isolated, concentrated, and added to a S. aureus group-I fluorescence reporter strain, AH1677 (Figure A), to gauge AIP production. Based on prior studies of the native L. monocytogenes 6-mer AIP in S. aureus, if this signal was successfully produced in B. subtilis, the supernatant should inhibit agr in S. aureus group-I cells. Significant inhibition was indeed observed in the S. aureus group-I agr reporter, reaching up to 75% inhibition at the highest concentration of supernatant tested. LC-MS/MS analyses of supernatant were performed to verify AIP biosynthesis. The major peak in the Lm sender cell supernatant matched the retention time, molecular ion mass to charge ratio (m/z), and MS/MS fragmentation patterns of an authentic 6-mer AIP standard (Figures B and S5). The fragmentation pattern also matched that of supernatant from a wild type L. monocytogenes strain in a previous report. There have been differing reports on the structure of the native L. monocytogenes AIP, ,− and our results here are congruent with the 6-mer AIP serving as a native signal. A standard curve of synthetic L. monocytogenes 6-mer (Figure S4) was used to quantify the amount of AIP in sender cell supernatant (at 8 h), and the sender cells were found to be producing approximately 12 nM AIP. WT L. monocytogenes was found to produce only a ∼3-fold higher level of AIP (34 nM) relative to the Lm sender cells (at 8 h), demonstrating that the AIP production level in the Lm construct was more similar to the native host than observed for the Sa or Se constructs. These results indicate that the Lm sender cells are capable of robust native AIP biosynthesis. Characterization of the engineered B. subtilis strains containing the L. monocytogenes agr QS system. (A) Dose–response of Lm sender cell supernatant in a S. aureus group-I agr fluorescent reporter (AH1677). Data normalized to media (0% activity) and WT B. subtilis 168 supernatant (100% activity) controls. (B) Mirror plot of MS2 spectra comparing Lm sender cell supernatant (blue, RT = 7.18 min) compared to a synthetic L. monocytogenes 6-mer AIP standard (black, RT = 7.18 min). Labeled mass identities represent peaks identified in both samples. (C) Fluorescence readout of the Lm receiver and full cells with and without xylose induction, and upon the addition of 10 μM 6-mer AIP. One biological replicate, consisting of three technical replicates, was conducted for these experiments. (D) Dose–response of 6-mer AIP in Lm receiver cells, showing a dose-dependent increase in fluorescence upon addition of compound. Fluorescence normalized to growth and media controls. Activity not reported as percent activation in panel (C) or (D) due to the lack of basal AIP production in the receiver cells that is usually standardized as “100% activity.” (E) Dose–response of inhibitor KdCdM in Lm receiver cells competed against 10 μM 6-mer AIP (blue curve) or alone (black curve). Fluorescence normalized to media (0% activity) and 10 μM 6-mer (100% activity for blue curve) or DMSO (100% activity for black curve). (F) Dose–response of inhibitor KdCdM in Lm full cells, showing a dose-dependent decrease in fluorescence as compound concentration increases. Data normalized to media (0% activity) and DMSO (100% activity) controls. Error bars for each plot indicate the standard error of the mean (SEM) of at least three biological replicates tested in technical triplicate (n = 3), unless indicated otherwise. See Experimental section for full details of methods. Receiver cells containing the L. monocytogenes agrCA genes and the L. monocytogenes P3 promoter controlling fluorescence readout were generated and characterized (i.e., Lm receiver cells), again using similar approaches as introduced above. Control experiments testing the Lm receiver cells with and without xylose were performed, and no significant difference was observed in fluorescence upon addition of xylose (Figure C). However, when the native L. monocytogenes 6-mer AIP was added, a dose-dependent increase in fluorescence was observed (Figure C,D). Notably, the normalized fluorescence levels for the Lm receiver cells were significantly lower than those observed for Sa and Se receiver cells, by over an order of magnitude. The origins for these lower levels are unclear; nevertheless, activation by the native AIP caused a significant change in fluorescence relative to the vehicle control. As expected, the known L. monocytogenes peptide inhibitor KdCdM (structure in Figure ) showed minimal inhibitory activity alone in the Lm receiver cells, but demonstrated dose-dependent inhibition when the system was pre-activated with exogenous native AIP (Figure E). L. monocytogenes full cells were next developed containing the entire L. monocytogenes agrBDCA operon and P3-controlled fluorescence readout system (i.e., Lm full cells). In control experiments testing these full cells with and without xylose, a significant increase in fluorescence was observed upon the addition of xylose, indicating that native AIP was being produced and activating the full cells (Figure C). Again, overall fluorescence levels were significantly lower than those observed for the other B. subtilis full systems reported in this study. Addition of the L. monocytogenes 6-mer AIP to the Lm full cells did not result in a significant difference from the vehicle (Figure C), indicating that this system could not be further activated, like the Sa full cells and unlike the Se full cells above. However, addition of the inhibitor KdCdM resulted in a dose-dependent decrease in fluorescence, resulting in 54% inhibition at 10 μM (Figure F). These results demonstrate that the family of Lm sender, receiver, and full cells are functional and provide evidence that this B. subtilis agr platform can be applied to Gram-positive species outside of the Staphylococci. Application of an Engineered B. subtilis Strain as a S. aureus Biosensor Since the receiver cell systems above were all capable of sensing native AIPs, we were curious whether this system could be reconfigured to become a biosensor for pathogens producing AIPs. We selected the community-associated methicillin-resistant group-I S. aureus strain, LAC, as the target for biosensor development due to its high clinical relevance. , Our goal was to design a system that could sense S. aureus AIP-I and in response produce a product capable of killing S. aureus. For this purpose, we selected lysostaphin, a bacteriolytic enzyme that targets and degrades the cell wall of Staphylococcal species (and is highly effective against MRSA), − in view of prior work using this product in biosensor design (as introduced above). , Construction of the biosensor strain in B. subtilis (termed SaRC-LysT) was completed using Gibson Assembly and Phusion polymerase chain reaction. The lysostaphin gene was fused to an AmyE secretion tag (IDT) as previously reported by the Voigt lab and inserted in place of the GFP gene in the Sa receiver cells, resulting in a cell line that should be capable of producing and secreting lysostaphin under S. aureus group-I agr control (see Experimental section for details of plasmid construction). To characterize the activity of the SaRC-LysT construct, zone of inhibition (ZOI) and supernatant assays were conducted. A lawn of S. aureus LAC was grown on agar plates containing 0.1% xylose with 5 μL aliquots of SaRC-LysT or WT B. subtilis (control) spotted onto the agar. To ensure that observed effects of SaRC-LysT were due to lysostaphin expression and not plasmid-selection antibiotics, cultures of B. subtilis 168 and SaRC-LysT were pelleted and then washed with and resuspended in PBS to remove residual antibiotics (Figure S11). The construct expressing lysostaphin produced a substantially larger ZOI when grown on S. aureus compared to a S. aureus agr mutant (LAC Δagr; Figure A). This reduction in activity with the mutant supports an AIP-dependent mechanism of activation in the biosensor cells, with residual inhibition likely reflecting basal expression of lysostaphin. S. aureus groups-II, -III, and -IV (RN6925, MW2, and RN4850, respectively) were also plated and examined in this assay with SaRC-LysT, and yielded ZOIs mirroring results from the Sa receiver cells, where the presence of AIP-I and AIP-III activate our system (Figures B and D; ZOI quantification in Figure S12). We note that there is a slight ZOI when S. aureus LAC is grown in the presence of the WT B. subtilis strain (Figure A,B). This effect is most likely attributed to B. subtilis production of bacilysin, a membrane associated antibiotic with weak activity against S. aureus. Characterization of the Sa receiver cell lysostaphin (LysT) construct. (A) Zone of inhibition (ZOI) assays to determine susceptibility. (Left) Lawn of S. aureus LAC grown in the presence of Sa receiver cell LysT or B. subtilis 168 (right half and left half, respectively). (Middle) Lawn of S. aureus LAC Δagr grown in the presence of Sa receiver cell LysT construct or B. subtilis 168 (right half and left half, respectively). (Right) ImageJ quantification and analysis of ZOI data at left and middle. Error bars represent the standard error of the mean (SEM) of three biological replicates (n = 3). (B) From left to right: Lawns of S. aureus RN6925 (group-II), MW2 (group-III), and RN4850 (group-IV) were grown and subjected to ZOI assays as described in part A. (C) S. aureus LAC growth data at 20 h when supplemented with various Bacillus supernatants (listed on x-axis). Percentages refer to xylose composition in the overnight Bacillus cultures. Empty refers to the empty vector control. Error bars indicate the standard error of the mean (SEM) of five biological replicates, each performed in technical duplicate (n = 5). Since lysostaphin is also active against S. epidermidis, albeit at higher concentrations, we tested SaRC-LysT against S. epidermidis groups-I, -II, and -III. S. epidermidis is a normal part of the human microbiome, and an ideal biosensor strain would not respond to other natural flora. The native AIPs from S. epidermidis are actually known to inhibit the group-I S. aureus agr system, and ZOI assays showed that the SaRC-LysT construct was largely inactive against S. epidermidis (Figure S13), as expected. These results are significant as they demonstrate a biosensor cell line that can sense and respond to the AIPs of multiple S. aureus specificity groups yet is nonresponsive to a closely related species. To further study the effects of the product(s) secreted by the SaRC-LysT, we used a supernatant-based assay in which filtered supernatants from overnight B. subtilis 168 or SaRC-LysT were added to S. aureus LAC cultures and growth was monitored by measuring OD600 after 24 h. Only supernatants from SaRC-LysT strains significantly reduced S. aureus LAC growth, with estimated lysostaphin concentrations of 0.63 and 1.25 μg/mL, for strains induced with 1% xylose or 1% xylose + AIP-I, respectively (lysostaphin quantified by reference to a standard curve; Figures C and S14). We reasoned that the growth inhibition seen for S. aureus LAC treated with the supernatant from SaRC-LysT induced with 1% xylose alone (lightest blue bar in Figure C) could stem from background expression of agr components. Supporting this hypothesis, further ZOI analyses at various xylose concentrations showed zones of inhibition for S. aureus LAC treated with SaRC-LysT supernatant at 0% xylose and for S. aureus LAC Δagr treated with SaRC-LysT supernatant induced with 1% xylose, indicating some leaky expression in this engineered system (Figure S15). Summary and Conclusions This study describes the development of a suite of engineered B. subtilis strains to sense and study the agr QS systems of Gram-positive bacteria. We generated (1) sender cells that contain the AIP biosynthesis components (i.e., the transmembrane peptidase AgrB and propeptide AgrD) and (2) receiver cells that contain the signal transduction components (i.e., the transmembrane receptor histidine kinase AgrC and response regulator AgrA) and a fluorescence reporter for AgrA activity. We first established the S. aureus agr system in B. subtilis as a proof-of-concept and then expanded the approach to the agr systems in a closely related species, S. epidermidis, and a more distant species, L. monocytogenes. We demonstrated that the Sa and Lm sender cells produced their expected AIPs (Sa AIP-I and 6-mer AIP, respectively) using both agr reporter assays on supernatant samples and LC-MS detection methods. While Se sender cell supernatants also were capable of agr inhibition in a reporter system, the Se AIP-I and/or miscleavage products were produced at insufficient levels for LC-MS detection. Full cell systems containing both AgrB/D and AgrC/A were then created for all three species and found to self-activate upon induction with xylose, indicating that the three agr operons were each functional in this heterologous host. This study represents, to our knowledge, the first time the S. epidermidis and L. monocytogenes agr QS circuits were examined outside of their native hosts, and the first time the signal generation and signal transduction components of the S. aureus agr circuit were studied in isolation and together systematically in B. subtilis. To further extend the scope of this study, we also reconfigured the Sa receiver system into a biosensor capable of sensing S. aureus AIP-I and producing the antibacterial enzyme lysostaphin upon detection at sufficient levels to kill S. aureus on agar plates and in liquid culture. This biosensor was able to detect three groups of S. aureus (-I, -III, and -IV) yet was insensitive to S. aureus group-II and S. epidermidis, highlighting its selectivity and potential utility. This study yielded several important outcomes. A first outcome is our demonstration that the engineered receiver and full cells are responsive to not only their native signals, but also a range of synthetic QSMs, with activity profiles that generally match those in wild-type hosts. This work demonstrates important proof-of-concept for the application of heterologous agr systems for the evaluation of exogenous compounds. The ability to isolate the agr system from other regulatory networks in the native host organismboth known and unknown to intersect with agris advantageous for mechanistic studies. Further, as B. subtilis is nonpathogenic and a BSL1 organism, the manipulation of these strains has obvious practical benefits relative to working with the host pathogens with regards to biosafety. We believe this system could be particularly useful for the study of putative agr components from other, less-studied Gram-positive bacteria to begin to characterize their QS circuits. Second, the B. subtilis heterologous strains provided insights into the underlying mechanisms of agr QS systems and of QSM activity, along with highlighting potential interesting differences between species. For example, the low level of AIP signal produced by Se sender cells relative to Sa and Lm sender cells indicates potential differences in AgrD processing when expressed in B. subtilis, which may also reflect variance in AIP biosynthesis between S. aureus, S. epidermidis, and L. monocytogenes. In turn, analysis of the AIP generated in Lm sender cells provided additional evidence for the 6-mer being a native AIP signal used by L. monocytogenes, a question of recent intrigue. When testing QSMs in the receiver constructs, unusual activation activity was observed for the inhibitors Sa AIP-III and D4A in the Sa receiver cells, and for the inhibitors Se AIP-III and AAA in the Se receiver cells. This activity in S. aureus is congruent with the results of a prior study by the Muir lab that showed unexpected activation for similar peptidic inhibitors in an in vitro AgrC-I activity assay. In our B. subtilis system, some inhibitory activity can be recovered for certain inhibitors when tested in competition with the native AIP, as demonstrated in the Se receiver and full cells. These discrepancies in activity could be attributed to partial agonism profiles in the B. subtilis system that are not observed in the native system. At least in the Se systems, compounds may also interact differently with AgrC when native AIP is present, a phenomenon that we have observed before for certain S. epidermidis agonists in the native host. Further experiments are necessary to investigate these differences, including in vitro experiments involving the S. epidermidis AgrC receptor and investigation of the inverse agonism trends observed with certain inhibitors in the S. aureus system. As a third outcome, we demonstrate that the B. subtilis receiver system can be readily redesigned to provide a S. aureus biosensor strain that is responsive to multiple native AIPs and generates the bacteriolytic agent lysostaphin at sufficient levels for killing of S. aureus. Although these results represent meaningful advances over previous agr biosensor work, our characterization of the system is still at an early stage. Additional experimentation to reduce leaky expression and determine sufficient levels of AIP needed for activation should further our understanding of the utility of this sensor system. Nevertheless, we note that a range of genes could be inserted after the P3 promoter in this receiver system, allowing for the AIP-dependent production of a wide variety of functional gene products depending on desired outcome. In addition, this biosensing system could be readily generalized to sense other agr-active bacteria beyond S. aureus. Overall, this suite of B. subtilis systems represent valuable biological tools that complement the growing arsenal of chemical modulators of QS and may offer new pathways for the study of cell–cell communication networks in Gram-positive bacteria. Supplementary Material