Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Open Access

Peer-reviewed

Research Article

Engineering a “detect and destroy” skin probiotic to combat methicillin-resistant Staphylococcus aureus

  • Changhui Guan ,

    Contributed equally to this work with: Changhui Guan, Peter J. Larson

    Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing – review & editing

    Affiliation The Jackson Laboratory for Genomic Medicine, Farmington, Connecticut, United States of America

    ⨯
  • Peter J. Larson ,

    Contributed equally to this work with: Changhui Guan, Peter J. Larson

    Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review & editing

    Affiliations The Jackson Laboratory for Genomic Medicine, Farmington, Connecticut, United States of America, UCONN Health (University of Connecticut), Farmington, Connecticut, United States of America

    ⨯
  • Elizabeth Fleming,

    Roles Data curation, Formal analysis, Investigation, Methodology

    Affiliation The Jackson Laboratory for Genomic Medicine, Farmington, Connecticut, United States of America

    ⨯
  • Alexander P. Tikhonov,

    Roles Formal analysis, Investigation, Methodology, Validation, Writing – review & editing

    Affiliation Azitra, Inc., Branford, Connecticut, United States of America

    ⨯
  • Sara Mootien,

    Roles Data curation, Investigation, Methodology, Validation, Writing – review & editing

    Affiliation Azitra, Inc., Branford, Connecticut, United States of America

    ⨯
  • Trudy H. Grossman,

    Roles Conceptualization, Data curation, Investigation, Methodology, Project administration, Supervision, Writing – review & editing

    Current address: CARB-X, Boston University School of Law, Boston, Massachusetts, United States of America

    Affiliation Azitra, Inc., Branford, Connecticut, United States of America

    ⨯
  • Caroline Golino,

    Roles Methodology

    Affiliation UCONN Health (University of Connecticut), Farmington, Connecticut, United States of America

    ⨯
  • Julia Oh

    Roles Funding acquisition, Project administration, Resources, Supervision, Writing – original draft, Writing – review & editing

    julia.oh@jax.org

    Affiliation The Jackson Laboratory for Genomic Medicine, Farmington, Connecticut, United States of America

    ⨯

Abstract

The prevalence and virulence of pathogens such as methicillin-resistant Staphylococcus (S.) aureus (MRSA), which can cause recurrent skin infections, are of significant clinical concern. Prolonged antibiotic exposure to treat or decolonize S. aureus contributes to development of antibiotic resistance, as well as depletion of the microbiome, and its numerous beneficial functions. We hypothesized an engineered skin probiotic with the ability to selectively deliver antimicrobials only in the presence of the target organism could provide local bioremediation of pathogen colonization. We constructed a biosensing S. epidermidis capable of detecting the presence of S. aureus quorum sensing autoinducer peptide and producing lysostaphin in response. Here, we demonstrate in vitro activity of this biosensor and present and discuss challenges to deployment of this and other engineered topical skin probiotics.

Citation: Guan C, Larson PJ, Fleming E, Tikhonov AP, Mootien S, Grossman TH, et al. (2022) Engineering a “detect and destroy” skin probiotic to combat methicillin-resistant Staphylococcus aureus. PLoS ONE 17(12): e0276795. https://doi.org/10.1371/journal.pone.0276795

Editor: D. Ashley Robinson, University of Mississippi Medical Center, UNITED STATES

Received: July 12, 2022; Accepted: October 13, 2022; Published: December 15, 2022

Copyright: © 2022 Guan et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: All relevant data are within the manuscript and its Supporting information files.

Funding: This project is supported by the Department of Defense W81XWH-18-1-0229 and National Institutes of Health (DP2 GM126893-01, K22 AI119231-01, and 1 R21 AR075174). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: JO is on the scientific advisory board of Azitra, Inc., Dermbiont, Inc., and ELSI Skin Health, Inc. AT, SM are employed by Azitra, Inc. TG is formerly employed by Azitra, Inc. This does not alter our adherence to LOS ONE policies on sharing data and materials.

Introduction

The human skin plays host to diverse commensal microorganisms with a broad array of contributions to skin function, immune education, wound repair, and even protection against neoplasia [1]. Clinically, this is often overshadowed by its role as a reservoir for pathogens [2], prompting efforts to decolonize the skin with broad-spectrum topical antibiotics at the cost of more beneficial commensals. In particular, Staphylococcus (S.) aureus is a major human pathogen causing a wide range of skin and soft tissue infections, as well as a leading cause of bacteremia and device-related infections [3]. Methicillin resistant S. aureus (MRSA) is widely disseminated in both healthcare and community settings across 6 continents [4, 5]. An estimated 30% of the US general population is colonized by S. aureus and 2% with MRSA [6], with prevalence considerably higher (up to 60%) among healthcare workers [7]. The wide dissemination of MRSA constitutes a major loss in the war on antimicrobial resistance, limiting use of beta-lactams and their derivatives as first-line antibiotics for potential staphylococcal infections.

Skin colonization by S. aureus is a major public health challenge because it is considered to increase risk of infection and infection recurrence [8–10]. Decolonization is typically attempted with topical antiseptics (e.g., chlorohexidine, triclosan) and/or topical antibiotics (e.g., mupirocin), which have the risk of incomplete decolonization together with depletion of the local microbiota, which could further increase re-infection risk [11]. Recolonization rates are as high as 60% [11] following such procedures, and biocides such as triclosan have even been shown to promote S. aureus colonization [12]. Antibiotic-induced depletion of commensal microbiota, whose diversity has been shown to improve cutaneous health, resist pathogen colonization, and bolster host immune response to pathogens [2, 13], may be an added inducement for alternative strategies to manage MRSA colonization.

Following the model of recent development of engineered probiotics to combat enteric infections by Pseudomonas [14] and Salmonella [15] pathogens, we sought to develop a skin probiotic to combat MRSA colonization and growth on skin. Ideally, we envisioned a probiotic capable of specifically targeting MRSA and while sparing beneficial commensal microbiota. We engineered S. epidermidis, a ubiquitous skin commensal, to express anti-MRSA antimicrobials under control of a S. aureus quorum sensing regulator. This “detect and destroy” system intends to deploy of antimicrobials only where and when S. aureus is present, allowing for local bioremediation. Here, we present the in vitro activity and in vivo efficacy of our engineered skin probiotic and discuss ongoing challenges in skin probiotic development.

Materials and methods

Ethics approval and consent to participate

All mouse experiments were approved by the Jackson Laboratory Animal Care and Use Committee. Mice were euthanized via CO2 narcosis.

Strains, plasmids and reagents.

All strains, plasmids and oligonucleotides (Table 1) used in this study are provided in the Supplementary Materials. Escherichia coli (E. coli) strains used for DNA cloning and sequencing were cultured in Lysogeny Broth (LB) (Difco Bacto) with or without 1.5% (w/v) agar. Plasmids were transformed into E. coli DH5α or E. coli dam-/dcm- chemically competent cells (New England Biolabs) according to the manufacturer’s protocol. Staphylococci were cultured in tryptic soy broth (TSB, BD Bacto) with or without 1.5% (w/v) agar. Antibiotics were added to the medium when needed: ampicillin (100 μg/mL), chloramphenicol (10–20 μg/mL), kanamycin (25–50 μg/mL) with concentrations depending on the host. Electrocompetent S. epidermidis ATCC12228 or Tü3298Δagr was prepared according to previously published methods [16] with modifications; electroporation was performed with a Gene Pulser Xcell Electroporation Systems (Biorad), using manufacturer-provided protocol with the following modifications. 0.5–1 μg of plasmid DNA was incubated with competent cells for 30 min at room temperature. Electroporation was performed at 2,300 volts, 25 μF, 100 Ω with 2 mm cuvettes. Plasmid DNA was prepared using Plasmid DNA Miniprep or Midiprep Kit (Qiagen). Sanger sequencing services were provided by Eurofins or Genewiz. Precast 10–20% tricine protein gels (Novex) were used to perform SDS-PAGE. Primers for polymerase chain reaction (PCR) or sequencing were ordered from Eurofin Scientific or Integrated DNA Technologies (IDT). Cloning of all constructs was conducted in E. coli DH5α. Verified constructs were transformed into E. coli dam-/dcm- competent cells to generate intermediate plasmids for S. epidermidis transformation.

thumbnail
Download:
Table 1. Primer sets used for amplifying specific DNA fragments by PCR.

https://doi.org/10.1371/journal.pone.0276795.t001

Construction of staphylococcal biosensor reporters.

S. aureus strains USA300-0114 (NR-46070, type I), NR-46204 (type II), NR-46081 (type III) and NR-45955 (type IV) were purchased from BEI Resources (www.beiresources.org). Primers for amplification of agrCA from the different S. aureus subtypes were designed using publicly available sequence data for these strains (GenBank Accession numbers: NR-46070-CAIHCM000000000.1, NR-46204-CAIHDX000000000.1, NR-46081- CP026066.1, NR-45955- CP026067.1). S. aureus genomic DNA was isolated with GenElute™ Bacterial Genomic DNA Kits from Sigma-Aldrich. agrCA and bidirectional promoter P2/P3 [17] were PCR amplified and ligated to plasmid pCN34 [18] sequentially. GFP gene was amplified from plasmid pCN57 [18] and ligated downstream of the P3 promoter in pCN34 and plasmid constructs were propagated into E. coli DH5a. For S. epidermidis genomic integration of the reporter constructs, the biosensor reporter cassette was cloned into temperature-sensitive E. coli-S. aureus shuttle vector pJB38 (gift from Jeffrey Bose), between two 1kb “arms” of the genomic sequences flanking the agr operon in S. epidermidis ATCC12228 for homologous recombination. Integration was performed according to methods previously published by Bose et al. [19].

Induction of biosensor reporters on plasmids in S. epidermidis.

To test the specificity and cross-reaction of the agrCA-P2/P3-GFP reporter constructs, 1:10 dilutions of filter-sterilized supernatants from overnight cultures of the four agr subtypes of S. aureus (USA300/agr type I, NR46204/agr type II, NR46081/agr type III, NR45955/agr type IV) were co-incubated with S. epidermidis Tü3298Δagr (1:100 dilution of fresh overnight cultures) harboring reporter plasmids of different subtypes in a total volume of 200 μL at 37°C for 5 hours with shaking in a 96-well plate (flat clear bottom and black sides, Cellstar). The cultures were centrifuged and washed twice with PBS and resuspended in 200 μL PBS, and the cell density (OD600) and fluorescence were read in BioTek Synergy2 plate reader.

Induction of chromosomal biosensor reporters in S. epidermidis.

To test the sensitivity of each biosensor construct on the chromosome, genomic integrants of biosensor reporters (integrated at the agr locus, resulting in a knockout of the locus) for agr types I, II and III in S. epidermidis ATCC12228 were co-incubated with dilutions (1:1, 1:10, 1:100, and 1:1000) of filter-sterilized supernatants of S. aureus of their corresponding agr subtypes in the same way as above.

Construction and induction of antimicrobial biosensors.

For antimicrobial peptide (AMP) expressing biosensor constructs, coding sequences for LL-37 (37 amino acids (AA)) and elafin [20] (61 AA) from human were synthesized by joint PCR with codons optimized for Staphylococcus. The gene for the bacteriocin hiracin (74 AA) from Enterococcus hirae [21] was PCR-synthesized without codon recoding. A gene encoding lysostaphin [22] (389 AA, S. simulans) was cloned from plasmid pDF8 in Bacillus subtilis BD170 (purchased from ATCC). The lysostaphin immunity gene was synthesized by joint PCR. To facilitate secretion of the antimicrobial products, a truncated signal peptide (MKFVKAIWPFVAVAIVFMFMSAA) derived from S. epidermidis secA secretion system [23] was cloned in front of the antimicrobial genes, except for the lysostaphin clone which contained an endogenous export motif [24]. A hexahistidine tag was added to the C-terminus of each AMP for purification.

The constructs were first cloned in plasmid pPL18 (Oh lab custom-made shuttle vector encoding kanamycin resistance, KanR) and the resulting plasmids (pPL18-AMPs) were transformed into S. epidermidis Tü3298 Δagr for activity and inducibility test. Indicator strain S. aureus RN4220/pCN34 (plasmid pCN34 carries a KanR gene) was plated on TSB agar plates containing 25 μg/mL of kanamycin and different concentrations of anhydrotetracycline (ATc, 0.5, 1.0, 1.5, 2.0, 2.5, 3.0, 3.5 and 4.0 μM) [25]. Ten microliters of overnight cultures of S. epidermidis Tü3298Δagr harboring pPL18-AMPs were spotted on top and the plates were incubated at 37°C overnight, then at room temperature for two more days for activity scoring.

To verify the expression and secretion of the antimicrobial peptides, one hundred milliliters of the bacterial cultures of S. epidermidis Tü3298Δagr containing pPL18-AMPs were induced at OD600 ~1.0 with 1.0 μM of ATc at 30°C for 4 hours. Batch purification of AMPs was carried out as follows: Two percent (v/v) of 1 M Tris-HCl, pH 8.0, and 10% (v/v) Ni-NTA resin (Qiagen) were added to the cell-free supernatant which was stirred at 4°C overnight to facilitate resin-peptide binding. The resin was collected in centrifuge tubes the next day and washed with washing buffer (20 mM Tris, 500 mM NaCl, 20 mM imidazole, pH 8.0). The bound peptide was eluted with elution buffer (20 mM Tris, 500 mM NaCl, 250 mM imidazole, pH 8.0). The eluate was dialyzed against phosphate buffered saline (PBS) and concentrated with Amicon spin columns (pore size 3 KD) with phosphate buffered saline (PBS). The samples were then subjected to SDS-PAGE analysis.

Integration and induction of biosensor antimicrobial peptide (AMP) constructs.

The antimicrobial biosensor constructs for agr type I, II, and III expressing lysostaphin were created by replacing the GFP gene with lysostaphin and its immunity genes in appropriate biosensor reporter constructs in the integrative plasmid pJB38. They were integrated via homologous recombination into the genome of S. epidermidis Tü3298 Δagr at the agr locus. Integration and DNA sequences were verified by sequencing the genomic PCR products. S. aureus of four agr types were plated on TSB agar plates and 10 μL containing 104 and 106 CFU of each of two individual integrants for each agr type were spotted on top. Parent strain S. epidermidis Tü3298 Δagr was included as control. The plates were incubated at 37°C overnight, then at room temperature for two more days for activity scoring.

Mouse trials.

Germ-free C57BL6/J mice were purchased from The Jackson Laboratory, Bar Harbor, ME and maintained at room temperature in a germ-free facility at UConn Health. To prepare bacterial inocula, overnight cultures (OD600 for all ~13) of S. aureus USA300, S. epidermidis Tü3298Δagr (parent) and S. epidermidis type I biosensor producing lysostaphin (biosensor) were centrifuged and the pellets were washed twice with sterile H2O. Groups of five mice were used, except where otherwise noted. The pellets of each individual strain or S. epidermidis-S.aureus mixtures were suspended in ~ 1.5 mL of Vaseline brand petroleum jelly [26], and applied as described below. For CFU quantitation, samples were collected by PurFlock Ultra buccal swabs (Puritan Medical Products) moistened in buffered solution (50 mM Tris-HCl, 1 mM EDTA and 0.5% Tween-20, pH 8.0), swabbing both ears for each sampling, and CFU quantitation was performed by spread plating onto mannitol salt agar (MSA) to differentiate S. aureus from S. epidermidis colonies. After incubation at 37°C overnight, colonies were counted, and identification of select colonizes were verified by MALDI-TOF. Colony counts over 2,500 were estimated.

For experiments examining growth dynamics after co-colonization, mice were inoculated singly with ~109 CFU of S. aureus USA300, S. epidermidis parent, or biosensor, or a mixture of S. aureus/S. epidermidis parent or biosensor (~107/109 CFUs, respectively). We found that colonization with a single strain was robust and relatively stable over the course of the experiment, with no significant differences between the three strains (Fig 3A). Each mouse was inoculated on both sides of both ears with the appropriate bacteria-petroleum jelly suspension on days 0, 2 and 4, described below. Beginning on day 2 and every other day through up to day 10, staphylococcal colonization was quantified as described above, prior to bacterial application.

For experiments examining growth dynamics after precolonization by S. epidermidis, mice were first colonized with ~109 CFUs parent or biosensor on day 0. On day 2, the mixture of S. aureus/S. epidermidis parent or biosensor (~107/109 CFUs) was applied. For one trial, an additional dose of ~109 CFUs parent or biosensor was applied on day 4, with negligible effect. Beginning on day 2 and every other day through up to day 10, staphylococcal colonization was quantified as described above prior to bacterial application.

For wound infection assays, germ-free mice were shaved, and then 4mm punch biopsies were performed to create wounds in the dorsal skin. A mixture of S. aureus/S. epidermidis (~105/108 CFU) in petroleum jelly was then inoculated into the wound. After 48 hours, a 1.5 cm x 1.5 cm area surrounding the wound was excised and harvested in PBS buffer. Bacteria were dissociated from skin in the PBS suspension by applying motorized Pellet Pestle (Fisher) disruption for 10 seconds, and dilutions were spread on MSA plates for CFU quantitation. n = 5 mice per group in initial trial, n = 12 for repeat trial.

Data analysis and statistics.

All data analysis was conducted in R (v4.0.2). Statistical significance in this study was attributed to tests wherein p < 0.05. Induction or suppression of plasmid biosensor fluorescent reporter by different AIPs (e.g., Fig 1D) was determined with a bidirectional T-test comparing to fluorescence of the TSB treatment group for the media sensor. Induction of integrated reporter constructs (e.g., Fig 1E) was assessed using an analysis of variance (ANOVA) with a post-hoc Tukey HSD test to assign groups. Linear regressions were performed using the Spearman method. Comparison of CFU counts over time between groups in mouse trials (Fig 3B-3E) was performed with analysis of co-variance (ANCOVA). Comparison of CFU burden in wound assays (Fig 3F-3G) was performed with a bidirectional T-test.

thumbnail
Download:
Fig 1. Biosensor design.

A) Illustration of the staphylococcal agr quorum-sensing circuit. AgrD pro-peptide is modified and exported by AgrB as an auto-inducer peptide (AIP). AIP stimulates two component sensor AgrCA, which upregulates transcription of the P2/P3 promoter, upregulating agr component and RNAIII expression. B) Concept illustration of a S. epidermidis “detect and destroy” probiotic biosensor. The sensing component from S. aureus’ quorum sensing circuit in A) (agrA and agrC) are introduced into an agr mutant S. epidermidis strain. Binding of AgrA to the P2/P3 promoter controlling expression of a S. aureus-targeting antimicrobial then results in density-dependent production of the antimicrobial of choice, here, lysostaphin. C) Staphylococcal biosensor circuit design for validation. agrCA sensor genes were cloned from S. aureus agr groups I-IV and placed downstream of promoter P2, with GFP under control of P3. D) Biosensor induction by different AIPs from different S. aureus agr types, testing the efficacy and the cross-reactivity of a plasmid-borne biosensor. Supernatants of overnight S. aureus cultures from each agr type were filter-sterilized, diluted 1:10 in TSB and co-incubated with S. epidermidis strains expressing the corresponding agr biosensor reporter constructs (sensor). Control was empty vector (pCN34). Bidirectional T-test compared to TSB control group, *: p ≤ 0.05, **: p ≤ 0.01, ***: p ≤ 0.001, n = 2 replicates. E) Testing efficacy and sensitivity of biosensor induction by the matched agr type AIP, using a biosensor integrated into the S. epidermidis genome. Given low cross-reactivity from D), we tested each biosensor with its matching agr type in a dilution series to test sensitivity. Biosensors were co-incubated with dilutions of supernatant from S. aureus strains of the matching agr type. Media was used as control. Significance between groups determined within each panel by ANOVA with post-hoc Tukey HSD test, p < 0.05, n = 3 replicates. Groups that do not share the same letter (a, b, c, or d) are significantly different in post-hoc multiple comparisons test.

https://doi.org/10.1371/journal.pone.0276795.g001

Results

Biosensor design

First, we selected S. epidermidis as the background for our engineered probiotic because it is a ubiquitous skin commensal [27]. Its widespread ability to colonize skin was desirable as we hypothesized that it might be able to then protect against MRSA anywhere on the body. Although relatively challenging to genetically modify, it is still among the most tractable of skin commensals [19].

Second, we desired that it would secrete a narrow-spectrum but anti-MRSA effector, and only do so in the presence of S. aureus. We reasoned that this was important because constitutive secretion of potentially toxic compounds could have undesired effects on skin and surrounding microbiome as well as potentially promote acquisition of resistance. Our design of biosensing co-opts S. aureus quorum sensing, a density-dependent mechanism by which many pathogens regulate virulence on a population scale by secreting chemicals called autoinducer peptides (AIPs) that accumulate extracellularly until a threshold concentration is achieved to trigger an alternative transcriptional program [28]. S. aureus AIP, which is species-specific [17], presumably could be detected by a S. epidermidis strain that possesses the corresponding signal transduction cascade.

In S. aureus, quorum sensing is controlled by the agr (accessory gene regulator) circuit [28]. Precursor peptide AgrD is modified and secreted by AgrB as an auto-inducer peptide (AIP). AgrC and AgrA form a two-component signaling system which upregulates the transcription of promoters P2 and P3 in response to AIP binding [17]. For expression in S. epidermidis, we cloned the agrCA sensor and P2/P3 promoter from MRSA strain USA300, assembled them into a staphylococcal shuttle vector such that agrCA remained under P2, and placed a GFP reporter gene downstream of P3.

In addition, because four different agr groups have been identified in S. aureus, each representing different phylogenetic groups [29], we developed sensors targeting all four groups. agrA and P2/P3 are highly conserved (98–100% nt identity) between agr groups, so we built derivative constructs by substituting the USA300 agr type I agrCA with agrCA cloned from representative strains of type II (derived from type II S. aureus strain NR46204), type III (NR46081), and type IV (NR45955). Finally, we expressed these plasmid constructs in S. epidermidis strain Tü3298Δagr [30]. An agr deletion mutant was used to avoid recombination events between high sequence identity genes, induction of the reporter gene by endogenous agr signaling, and to circumvent potential cross-inhibition between S. epidermidis and S. aureus agr circuits [17, 31], which could cause this biosensor to block itself. The overall design is shown in Fig 1.

To test the activation of these prototype biosensors, we examined the response to AIP from its own group as well as cross-reactivity between groups (Fig 1D). Here, we used a plasmid-borne biosensor, incubating clones of each biosensor type with a 1:10 dilution of culture supernatant from self or other agr types. As expected, supernatant from other agr subtypes had variable effect on biosensor induction. The most potent induction (>2-fold increase in fluorescence over media control) typically occurred with its matched type (e.g., type I supernatant with type I biosensor), and cross-reactivity of the type I and type II biosensors was low. Interestingly, the type III biosensor had similar strong activation from both type III and type IV supernatant, which may be consistent with the consideration of agr type IV as a rare transitional genotype, having properties that are similar to multiple agr types [32]. Finally, the type IV biosensor had low activation by any supernatant including itself and had cross-reactivity with the type III strain. Because type II and type III agr biosensors were able to detect type IV AIP more effectively than the type IV biosensor itself, and type IV agr strains have also been shown to have weak cross reactivity with agr type I AIP [32], we excluded the type IV construct from subsequent experiments. We note that the observed differences in cross-interactions were not entirely consistent with published interactions between agr groups [29, 32], and could arise from the heterologous expression setting, or other elements of the supernatant such as bacteriocins, extracellular enzymes, or pH differences.

We noted that our plasmid-borne biosensor had relatively high background GFP expression (Fig 1D), likely due to low constitutive expression from the P2/P3 promoter combined with the moderate copy number of the plasmid. Given that our ultimate goal is to express toxins or antimicrobial peptides (AMPs), it would be ideal to have a system with lower basal expression. We hypothesized that copy number could contribute to high background expression (though vector pCN34 is characterized as a low copy staphylococcal plasmid [18]), so we integrated the type I-III biosensor constructs into the chromosome at the residual agr locus and tested activation and cross-reactivity as described above. We observed significantly reduced background GFP expression and a dose-dependent induction, with induction occurring most potently at 1:1 or 1:10 dilutions, albeit with a reduction in overall expression (Fig 1E). While this is a tradeoff to minimizing background expression that may reduce biosensor efficacy, we deemed it necessary, and it had the added benefit that no agents were subsequently needed to maintain plasmid-borne transgenes, which could present additional challenges to deploy in vivo.

Having confirmed that gene expression in S. epidermidis could be controlled by S. aureus quorum signaling in vitro, we next sought to determine what antimicrobials might be a feasible output. Our major considerations were 1) effective inhibition of clinically important S. aureus strains such as USA300, 2) potential to encode self-immunity in the S. epidermidis biosensor, and 3) feasibility of heterologous expression (i.e., larger constructs might be more difficult to deploy). Our initial trials included epidermin, a native, plasmid-borne S. epidermidis lantibiotic with a broad spectrum of activity that is readily expressed heterologously with built-in self-immunity [33]. However, it had no inhibitory activity against most S. aureus strains (not shown). We then attempted dual plasmid systems to pre-express resistance cassettes to lantibiotics gallidermin [33] and mersacidin [34], then introduce the synthesis genes in a second vector. This approach had limited success, likely due to the size of the operon or transgene toxicity. Given these challenges, we adjusted our strategy to focus on small peptides or proteins that could be produced with a single heterologous gene. We compiled a library of ~200 bacteriocin genes/peptides from the literature and relevant databases (e.g., Uniprot at https://www.uniprot.org/ [35], Bactibase at http://bactibase.hammamilab.org/ [36]) and evaluated these for source, size, self-immunity, post-translational modifications required, requirement for transporters, potential host resistance, and activity against S. aureus. We identified several strong candidates, including human AMPs LL-37 [37] and elafin [20], E. hirae bacteriocin hiracin [21], and S. simulans bacteriolytic enzyme lysostaphin [22]. The former two were of interest given their size (<10 kDa) and potential tractability, despite that they would likely be toxic to the expression host, and the latter two were of interest for their size (8 kDa and 42 kDa, respectively) and their potential for encoding expression host resistance. Each had demonstrated activity against S. aureus with no or less potency against S. epidermidis than had been previously reported, except for lysostaphin, for which we introduced a facilitator gene for resistance.

We first tested if these four candidates could be expressed heterologously in S. epidermidis. Because of potential toxicity, we tested them under a tightly controlled inducible tetR/Pxyl/tet promoter system [25] (Fig 2A). Self-immunity genes were included for hiracin and lysostaphin to prevent self-killing. To identify potential effective secretion signal tags for the antimicrobials, we screened a Gram + Sec-type signal peptide (SP) library [23], cloning them onto the 5’ end of LL-37 in E. coli as our test case. Out of 11 unique SPs identified in E. coli hosts, we were able to transform 9 into Tü3298Δagr. Of these, we performed a growth curve inhibition assay with the assumption that the more effectively LL-37 is secreted upon anhydrous tetracycline (ATc) induction, there would be less growth defect from the intracellular accumulation of toxic product. Two clones showed no changes in doubling time (SP-15 and SP-17) and one clone showed a slight slowing in doubling time (SP-14) after induction. All three were selected for proof-of-principle testing with LL-37 (Fig 2B), and SP-15 was used for the remaining candidates, except for lysostaphin, for which we used the endogenous signal peptide [24].

thumbnail
Download:
Fig 2. In vitro efficacy of S. epidermidis producing different antimicrobials.

A) Concept of design to express antimicrobial peptide (AMP) genes under the biosensor. AMPs are tagged for export with a signal peptide (SP) in S. epidermidis under control of either (1) tetracycline inducible promoter (Pxyl/tet) for testing under tight inducible control, or (2) agr P2P3 biosensor. B) Pull-down SDS-PAGE verifying AMPs elafin, LL-37, hiracin, and lysostaphin secretion from S. epidermidis. Several different signaling peptides (e.g., SP-14, SP-15, SP-17) were tested for LL-37. The endogenous signal peptide for lysostaphin was used. C) S. epidermidis expressing AMPs under the tet-inducible promoter demonstrate strong growth inhibition of S. aureus in an overlay assay. Concentration of anhydrotetracycline (ATc) is shown and is dose-dependent. D) In vitro activity and specificity of lysostaphin-expressing agr type I, II, and III biosensors integrated into Tü3298Δagr genome. Parent strain included as control. Two biosensor integrant clones (i.e., biological replicates “1” and “2”) were tested at initial doses of 104 and 106 CFUs of S. epidermidis spotted on for each overlay assay.

https://doi.org/10.1371/journal.pone.0276795.g002

We verified the expression and secretion of the antimicrobials by purification of the peptides from supernatant with a Ni-NTA pulldown assay and SDS-PAGE (Fig 2B). Then, to ascertain that the secreted peptides were active against S. aureus, isolates were grown in an S. aureus agar overlay assay with increasing concentrations of ATc for antimicrobial induction. Activity was scored by zone of inhibition (Fig 2C), with all antimicrobial-producing isolates showing potent activity against S. aureus with a linear dose-response to ATc. We noted that the activity of LL-37 with SP-14 was lower than the same versions with other signal peptides, likely due to a different sequence composition in the signal peptide, consistent with the observed growth defect. For example, with 3.0 μM of ATc, the inhibition zone of LL-37/SP-14 had a diameter of 14 mm, while LL-37 with other SPs had a diameter of 21 mm. Under the tested conditions, all AMPs showed dose-dependency: with ATc at 2.5, 3.0, and 3.5 μM, the diameters of the inhibition zones were consistent (with the exception of LL-37/SP-14) at 17, 21 and 28 mm, respectively. Taken together, we concluded that S. epidermidis was an effective producer of S. aureus antimicrobials irrespective of previously reported activity against S. epidermidis itself (i.e., LL-37 [37]).

Given comparable efficacy between these select antimicrobials, we proceeded with lysostaphin as our proof-of-principle in the biosensor construct, as it has the narrowest spectrum of activity given that its substrate is a staphylococcal peptidoglycan, and genes encoding resistance can also be cloned into the biosensor host. In addition, its minimum inhibitory concentration (MIC) as well as its bacteriocidal dynamics against S. aureus is well characterized (killing S. aureus within minutes and MIC at which 90% of the strains are inhibited [MIC90], 0.001 to 0.064 μg/ml [22]). We placed the lysostaphin-encoding locus under the type I-III biosensors and integrated the resulting constructs plus resistance gene into the Tü3298Δagr genome (Fig 2A). We sought to examine its activity in an agar overlay assay, which differs from the in vitro liquid assay as interactions of the biosensor strain with S. aureus is spatially defined. When incubated with agar overlays of S. aureus strains from each agr group, we observed corresponding zones of inhibition for each target group (Fig 2D). Activities for agr groups I, II, and III were agr group specific, with negligible cross-activation (Fig 2D). These results confirmed that S. epidermidis could be engineered to “detect and destroy” S. aureus in an agr group-specific manner in vitro, excluding agr group IV, which we found had similar cross-reactivity in this assay as well.

We next sought to determine whether a staphylococcal biosensor could inhibit MRSA colonization or proliferation in vivo using a gnotobiotic C57BL6J mouse model. Mouse ears were treated with petroleum jelly suspensions of the lysostaphin-expressing type I biosensor strain, its parent strain Tü3298Δagr as a negative control, type I S. aureus MRSA strain USA300, or a combination thereof. For all experiments (except where otherwise noted, see Fig 3 legend), inocula included 107 CFU of MRSA and 109 CFU of S. epidermidis. While these densities of bacteria are higher than estimated in human skin [38], they are typical of experiments applying human staphylococci to mouse skin [39]. Microbial mixtures were applied at 0, 2, and 4 days, and colonization of S. epidermidis or S. aureus was assayed by colony-forming unit (CFU) quantitation for up to 10 days. As a control, both S. epidermidis strains colonized adequately, consistent with our prior use of this model, and no significant fitness defect was noted in the biosensor compared to its parent strain (Fig 3A). We performed several different iterations and replicates of this experiment to model potential real-world situations, and to attempt to resolve high variance in the model. We altered: Fig 3B and 3C) dosage, Fig 3D and 3E) timing of colonization (concurrent colonization vs. pre-colonization by the biosensor), or Fig 3F–3G) changing the host environment by assessing colonization dynamics in breached skin (skin punch biopsy wound).

thumbnail
Download:
Fig 3. Trials to define biosensor control of MRSA proliferation in vivo.

Germ-free C57BL6/J mice were colonized on the ears on days 0, 2 and 4, and growth of S. epidermidis (agr type I lysostaphin producing biosensor or Tü3298Δagr parent strain) and S. aureus was quantified by swabbing the ear, directly spreading on mannitol salt agar, and then counting CFUs on days 2, 4, 6, 8, and 10 (select cases). In A) we show representative control samples for colonization alone by S. aureus USA300 (MRSA), S. epidermidis parent, or S. epidermidis biosensor strains. B) and C), D) and E), and F) and G) are replicate experiments. In B) and C), we examined biosensor efficacy in suppressing growth of MRSA when co-colonized. MRSA was mixed with the biosensor or parent in a suspension at a 1:100 ratio (107/109 CFUs, respectively) and then applied to the mouse ears. n = 5 mice for all groups. Spearman linear regression line is shown with 95% confidence intervals. P-value to assess statistical significance in CFU counts between groups, accounting for time as a covariate, was determined within each panel by ANCOVA. In D) and E), we examined biosensor efficacy in preventing colonization (and suppressing growth), colonizing mouse ears with biosensor or parent alone at day 0, then challenging with the suspension of S. aureus USA300 with S. epidermidis biosensor or parent on day 2. n = 5 for all groups. In F) and G), we examined growth in a wound model, in which mice underwent a dorsal punch biopsy which was then inoculated with the suspension of S. aureus USA300 plus biosensor or parent. After 48 h, dorsal skin surrounding the wound was harvested for CFU quantitation on mannitol salt agar. n = 5 mice/group (left), n = 12/group (right). P-value was determined by bidirectional t-test.

https://doi.org/10.1371/journal.pone.0276795.g003

Biosensor efficacy would, ideally, manifest as low/no increase in S. aureus CFUs over the course of the experiment, compared to the parent strain (slope = 0 vs. slope > 0, or CFUs biosensor << CFUs parent at endpoint, respectively). Although conducted rigorously with germ-free mice, we found that there was 1) effective but very high variance in colonization (or quantitation) of either staphylococci, both within- and between- cohorts/cage. We used a range of birth cohorts of younger mice (~1 month of age), and randomized and individually housed mice where possible, but could not identify a consistent confounder that could explain the variance. 2) Interestingly, we observed that the parent strain alone could be effective at blocking S. aureus growth, suggesting that colonization resistance could, to a degree, be imparted by physical occupation of the environmental niche. 3) S. epidermidis colonization also varied, and tended to decrease over time, while S. aureus typically increased (with or without biosensor), suggesting that this strain of S. epidermidis may not be well suited to growth on the mouse skin, which we believe could impact efficacy of antimicrobial production. Taken together, the numerous variables with the colonization model resulted in 4) variable biosensor efficacy from cohort to cohort, which we believe was a consequence of insufficient density-dependent production of S. aureus AIP, insufficient S. epidermidis cells to produce adequate antimicrobial to significantly reduce S. aureus growth, and/or environmental factors that otherwise affected S. epidermidis fitness and antimicrobial production. The gnotobiotic model (similar to results in an in vitro 3D reconstructed human epidermis model), suffered from high variance in colonization efficacy, which we believe impacts the underlying biology of our biosensor, which requires sufficient accumulation of S. aureus AIP for antimicrobial production.

Discussion

While far from comprehensive, these results provide a proof-of-concept that S. epidermidis can be engineered to detect and respond to S. aureus, express exogenous antimicrobial proteins or peptides under this system, and potentially control MRSA proliferation via these mechanisms. A similar S. aureus sensor mechanism was recently developed by Lubkowicz et al. in Lactobacillus reuteri, showing the impressive molar sensitivity of type I agrCA [40]. We conjectured that our use of S. epidermidis, which unlike lactobacilli, ubiquitously colonizes every human skin site [27], and placing antimicrobial genes as an output, would open the door to using this technology not just as a detector but also a potential prophylactic or therapeutic in the skin. While antimicrobials could be expressed constitutively in S. epidermidis, we also sought to circumvent potential fitness and auto-toxicity burdens to the host and potentially adverse disruptions to the native microbiota.

However, we found that while our biosensor performed impressively and reproducibly in vitro, in vivo efficacy was highly variable. We attribute this to numerous potential factors previously discussed, most likely, in our opinion, differential colonization efficacy, density, and ratio of S. epidermidis:S. aureus, as proximity is requisite for biosensor activation. We note that we were unable to ascertain density of staphylococcal colonization, or actual amount of AIP present that would be sufficient to activate the biosensor. In addition, we acknowledge that we performed only qualitative assessments of antimicrobial production, as we deemed in vivo efficacy to be the first bar to justify a rigorous quantitation as a function of colonization efficacy, density, and environmental conditions. While imaging or mass spectrometry presumably could be used to prove density as a variable, due to the longitudinal nature of the experiment and the difficulties in accurate inference, we did not attempt this. From experiments in reconstructed human epidermis, we have seen ‘patchiness’ in colonization with GFP-labeled strains that supports this hypothesis. In addition, recovery and quantitation by swabbing may impart additional variance given potential grooming by the animals or adherence of cells to the skin.

Similarly, antimicrobial activity is likely highly localized to S. epidermidis cells, as supported by our in vitro data. An additional control for future in vivo experiments might include a strain that constitutively produces the antimicrobial of interest (or our inducible version from 2C), to specifically evaluate the ability of the biosensor to sense then produce antimicrobial. In addition, we note that even when a trial could be deemed successful, MRSA was never entirely cleared. This may be an inherent disadvantage of deriving the biosensing mechanism from a quorum-sensing regulator, as these have evolved for population density-dependent signaling, rather than maximum sensitivity. A potential future direction for this technology could be to optimize the sensitivity of the AgrCA regulator with directed evolution. Higher sensitivity may facilitate a greater reduction in MRSA colonization, a more rapid suppression of its expansion, or importantly, given our in vivo results, a greater tolerance to low relative density of the biosensor with respect to its target. In addition, screening/engineering additional promoters for higher production of the antimicrobial may improve efficacy. Finally, other agr type biosensors should be characterized, in addition to other S. epidermidis strains, which may have different colonization dynamics. We pursued the type I biosensor as the most likely to be clinically pertinent, given the prevalence and virulence of agr type I strains [28]. As an aside, we extensively attempted to engineer an omnisensor (one cell capable of sensing all 4 types), using different regions of the agrA, agrC, and P2/P3 from all agr types. However, these efforts were not fruitful, likely because of high homology between the regions, despite introducing numerous synonymous base pair changes.

While we were encouraged by the strong in vitro performance of this “detect and destroy” probiotic approach, we feel that our inconclusive in vivo results reflect real-world challenges facing this and other topical probiotics. We and others established the biodiversity, interindividual, and intraindividual differences in the skin microbiome [13, 27, 41, 42]. Colonization of exogenous S. epidermidis may be hampered by numerous physiological considerations, including penetrance through the stratum corneum and colonization into deeper adnexal structures, differing physiological conditions of different skin sites (e.g., oiliness/moistness/dryness/pH), and/or competition against the endogenous microbiota, including other S. epidermidis strains which may have pre-adapted to the niches of various skin sites. In addition, it is possible that agr alters colonization efficacy, and we used an agr deficient S. epidermidis strain–reports on agr’s requirement for colonization short and long-term are conflicting, or have been assessed in different species [43–45]. Additional characterizations of S. epidermidis and S. aureus growth dynamics and interactions in vivo may be warranted for future applications to create an improved model that can account for these real-world differences.

Finally, an important consideration for genetically engineered microbial therapeutics are methods of biocontainment, preventing the transmission of the heterologous organism from the target site or from person to person, e.g., auxotrophy, foreign nucleotides and amino acid usage, and kill switches have been proposed or utilized [46, 47]. While understanding the underlying biology of colonization of exogenous strains will be a critical next step, we hope that this work provides an example of a potential initial design of a “detect and destroy” probiotic approach for mitigation of S. aureus infection.

Supporting information

S1 File. Raw data underlying figures.

https://doi.org/10.1371/journal.pone.0276795.s001

(XLSX)

Acknowledgments

We would first like to thank Michael Otto and his laboratory for their gift of the S. epidermidis Tü3298Δagr strain we used throughout this study. We are grateful to Jeffrey Bose and his laboratory for the gift of pJB38. The Jackson Laboratory Gnotobiotic Core, particularly Michael Michaud and Mayerlin Chalarca were instrumental implementing our mouse trials. Finally, we would like to acknowledge the other past and present members of the Oh lab for their contributions to this project and ongoing support.

References

  1. 1. Boxberger M, Cenizo V, Cassir N, Scola BL. Challenges in exploring and manipulating the human skin microbiome. Microbiome. 2021;9. pmid:34053468
  2. 2. Flowers L, Grice EA. The Skin Microbiota: Balancing Risk and Reward. Cell host & microbe. 2020;28: 190–200. pmid:32791112
  3. 3. Tong SYC, Davis JS, Eichenberger E, Holland TL, Fowler J, Vance G. Staphylococcus aureus infections: epidemiology, pathophysiology, clinical manifestations, and management. Clinical microbiology reviews. 2015;28: 603–661. pmid:26016486
  4. 4. Rasigade J-, Dumitrescu O, Lina G. New epidemiology of Staphylococcus aureus infections. Clin Microbiol Infect. 2014;20: 587–588. pmid:24930666
  5. 5. Monecke S, Coombs G, Shore AC, Coleman DC, Akpaka P, Borg M, et al. A field guide to pandemic, epidemic and sporadic clones of methicillin-resistant Staphylococcus aureus. PLoS ONE. 2011;6: e17936. pmid:21494333
  6. 6. Peterson J, Garges S, Giovanni M, McInnes P, Wang L, Schloss JA, et al. The NIH Human Microbiome Project. Genome Res. 2009;19: 2317–2323. pmid:19819907
  7. 7. Dulon Madeleine, Peters Claudia, Schablon Anja, Nienhaus Albert. MRSA carriage among healthcare workers in non-outbreak settings in Europe and the United States: a systematic review. BMC Infectious Diseases. 2014;14: 363. pmid:24996225
  8. 8. Al-Zubeidi D, Burnham CD, Hogan PG, Collins R, Hunstad DA, Fritz SA. Molecular Epidemiology of Recurrent Cutaneous Methicillin-Resistant Staphylococcus aureus Infections in Children. J Pediatric Infect Dis Soc. 2014;3: 261–264. pmid:25844165
  9. 9. Creech CB, Al-Zubeidi DN, Fritz SA. Prevention of Recurrent Staphylococcal Skin Infections. Infect Dis Clin North Am. 2015;29: 429–464. pmid:26311356
  10. 10. Kao K, Chen C, Hu H, Chang H, Huang C, Huang Y. Risk Factors of Methicillin-Resistant Staphylococcus aureus Infection and Correlation With Nasal Colonization Based on Molecular Genotyping in Medical Intensive Care Units: A Prospective Observational Study. Medicine (Baltimore). 2015;94: e1100. pmid:26181545
  11. 11. Price A, Sarween N, Gupta I, Baharani J. Meticillin-resistant Staphylococcus aureus and meticillin-susceptible Staphylococcus aureus screening in a cohort of haemodialysis patients: carriage, demographics and outcomes. J Hosp Infect. 2015;90: 22–27. pmid:25676114
  12. 12. Syed AK, Ghosh S, Love NG, Boles BR. Triclosan promotes Staphylococcus aureus nasal colonization. MBio. 2014;5: e01015. pmid:24713325
  13. 13. Byrd AL, Belkaid Y, Segre JA. The human skin microbiome. Nature Reviews Microbiology. 2018;16: 143. pmid:29332945
  14. 14. Hwang IY, Koh E, Wong A, March JC, Bentley WE, Lee YS, et al. Engineered probiotic Escherichia coli can eliminate and prevent Pseudomonas aeruginosa gut infection in animal models. Nat Commun. 2017;8: 15028. pmid:28398304
  15. 15. Palmer JD, Piattelli E, McCormick BA, Silby MW, Brigham CJ, Bucci V. Engineered Probiotic for the Inhibition of Salmonella via Tetrathionate-Induced Production of Microcin H47. ACS Infect Dis. 2018;4: 39–45. pmid:28918634
  16. 16. Löfblom J, Kronqvist N, Uhlén M, Ståhl S, Wernérus H. Optimization of electroporation-mediated transformation: Staphylococcus carnosus as model organism. Journal of Applied Microbiology. 2007;102: 736–747. pmid:17309623
  17. 17. Otto M, Echner H, Voelter W, Gotz F. Pheromone cross-inhibition between Staphylococcus aureus and Staphylococcus epidermidis. Infect Immun. 2001;69: 1957–1960. pmid:11179383
  18. 18. Charpentier E, Anton AI, Barry P, Alfonso B, Fang Y, Novick RP. Novel cassette-based shuttle vector system for gram-positive bacteria. Appl Environ Microbiol. 2004;70: 6076–6085. pmid:15466553
  19. 19. Bose JL. Genetic manipulation of staphylococci. Methods Mol Biol. 2014;1106: 101–111.
  20. 20. Simpson AJ, Maxwell AI, Govan JRW, Haslett C, Sallenave J-. Elafin (elastase-specific inhibitor) has anti-microbial activity against Gram-positive and Gram-negative respiratory pathogens. FEBS Letters. 1999;452: 309–313. pmid:10386612
  21. 21. Sánchez J, Diep DB, Herranz C, Nes IF, Cintas LM, Hernández PE. Amino acid and nucleotide sequence, adjacent genes, and heterologous expression of hiracin JM79, a sec-dependent bacteriocin produced by Enterococcus hirae DCH5, isolated from Mallard ducks (Anas platyrhynchos). FEMS Microbiol Lett. 2007;270: 227–236. pmid:17326750
  22. 22. Wu JA, Kusuma C, Mond JJ, Kokai-Kun JF. Lysostaphin Disrupts Staphylococcus aureus and Staphylococcus epidermidis Biofilms on Artificial Surfaces. Antimicrob Agents Chemother. 2003;47: 3407–3414.
  23. 23. Brockmeier U, Caspers M, Freudl R, Jockwer A, Noll T, Eggert T. Systematic Screening of All Signal Peptides from Bacillus subtilis: A Powerful Strategy in Optimizing Heterologous Protein Secretion in Gram-positive Bacteria. Journal of Molecular Biology. 2006;362: 393–402. pmid:16930615
  24. 24. Bastos Maria do Carmo de Freire, Coutinho BG, Coelho MLV. Lysostaphin: A Staphylococcal Bacteriolysin with Potential Clinical Applications. Pharmaceuticals (Basel). 2010;3: 1139–1161. pmid:27713293
  25. 25. Bertram R, Hillen W. The application of Tet repressor in prokaryotic gene regulation and expression. Microb Biotechnol. 2008;1: 2–16. pmid:21261817
  26. 26. Mohammad H, Cushman M, Seleem MN. Antibacterial Evaluation of Synthetic Thiazole Compounds In Vitro and In Vivo in a Methicillin-Resistant Staphylococcus aureus (MRSA) Skin Infection Mouse Model. PLoS One. 2015;10: e0142321. pmid:26536129
  27. 27. Zhou W, Spoto M, Hardy R, Guan C, Fleming E, Larson PJ, et al. Host-Specific Evolutionary and Transmission Dynamics Shape the Functional Diversification of Staphylococcus epidermidis in Human Skin. Cell. 2020;180: 454–470.e18. pmid:32004459
  28. 28. Le KY, Otto M. Quorum-sensing regulation in staphylococci-an overview. Front Microbiol. 2015;6: 1174. pmid:26579084
  29. 29. Otto M. Staphylococcus aureus and Staphylococcus epidermidis peptide pheromones produced by the accessory gene regulator agr system. Peptides. 2001;22: 1603–1608. pmid:11587788
  30. 30. Vuong C, Götz F, Otto M. Construction and characterization of an agr deletion mutant of Staphylococcus epidermidis. Infect Immun. 2000;68: 1048–1053. pmid:10678906
  31. 31. Thoendel M, Kavanaugh JS, Flack CE, Horswill AR. Peptide signaling in the Staphylococci. Chem Rev. 2011;111: 117–151. pmid:21174435
  32. 32. WrightIII JS, Traber KE, Corrigan R, Benson SA, Musser JM, Novick RP. The agr Radiation: an Early Event in the Evolution of Staphylococci. Journal of Bacteriology. 2005.
  33. 33. Götz F, Perconti S, Popella P, Werner R, Schlag M. Epidermin and gallidermin: Staphylococcal lantibiotics. International Journal of Medical Microbiology. 2014;304: 63–71. pmid:24119540
  34. 34. Altena Karsten, Guder André, Cramer Claudia, Bierbaum Gabriele. Biosynthesis of the Lantibiotic Mersacidin: Organization of a Type B Lantibiotic Gene Cluster. Applied and Environmental Microbiology. 2000;66: 2565–2571. pmid:10831439
  35. 35. The UniProt Consortium. UniProt: the universal protein knowledgebase in 2021. Nucleic Acids Res. 2020;49: D480–D489. pmid:33237286
  36. 36. Hammami R, Zouhir A, Lay CL, Hamida JB, Fliss I. BACTIBASE second release: a database and tool platform for bacteriocin characterization. BMC Microbiology. 2010;10: 22. pmid:20105292
  37. 37. Noore J, Noore A, Li B. Cationic Antimicrobial Peptide LL-37 Is Effective against both Extra- and Intracellular Staphylococcus aureus. Antimicrob Agents Chemother. 2013;57: 1283–1290. pmid:23274662
  38. 38. Skowron K, Bauza-Kaszewska J, Kraszewska Z, Wiktorczyk-Kapischke N, Grudlewska-Buda K, Kwiecińska-Piróg J, et al. Human Skin Microbiome: Impact of Intrinsic and Extrinsic Factors on Skin Microbiota. Microorganisms. 2021;9. pmid:33808031
  39. 39. Naik S, Bouladoux N, Linehan JL, Han S, Harrison OJ, Wilhelm C, et al. Commensal-dendritic-cell interaction specifies a unique protective skin immune signature. Nature. 2015;520: 104–108. pmid:25539086
  40. 40. Lubkowicz D, Ho CL, Hwang IY, Yew WS, Lee YS, Chang MW. Reprogramming Probiotic Lactobacillus reuteri as a Biosensor for Staphylococcus aureus Derived AIP-I Detection. ACS Synth Biol. 2018;7: 1229–1237. pmid:29652493
  41. 41. Oh J, Byrd AL, Deming C, Conlan S, Kong HH, Segre JA. Biogeography and individuality shape function in the human skin metagenome. Nature. 2014;514: 59–64. pmid:25279917
  42. 42. Oh J, Byrd AL, Park M, Kong HH, Segre JA. Temporal Stability of the Human Skin Microbiome. Cell. 2016;165: 854–866. pmid:27153496
  43. 43. Nakamura Y, Takahashi H, Takaya A, Inoue Y, Katayama Y, Kusuya Y, et al. Staphylococcus Agr virulence is critical for epidermal colonization and associates with atopic dermatitis development. Sci Transl Med. 2020;12. pmid:32641488
  44. 44. Tan L, Li SR, Jiang B, Hu XM, Li S. Therapeutic Targeting of the Staphylococcus aureus Accessory Gene Regulator (agr) System. Front Microbiol. 2018;0. pmid:29422887
  45. 45. Shopsin B, Drlica-Wagner A, Mathema B, Adhikari RP, Kreiswirth BN, Novick RP. Prevalence of agr dysfunction among colonizing Staphylococcus aureus strains. J Infect Dis. 2008;198: 1171–1174. pmid:18752431
  46. 46. Dodds D, Bose JL, Deng M, Dubé GR, Grossman TH, Kaiser A, et al, Controlling the Growth of the Skin Commensal Staphylococcus epidermidis Using d-Alanine Auxotrophy. mSphere. 2020;5. pmid:32522780
  47. 47. Pedrolli DB, Ribeiro NV, Squizato PN, de Jesus VN, Cozetto DA. Engineering Microbial Living Therapeutics: The Synthetic Biology Toolbox. Trends in biotechnology. 2019;37. pmid:30318171