*Article* **Discovery of** *Staphylococcus aureus* **Adhesion Inhibitors by Automated Imaging and Their Characterization in a Mouse Model of Persistent Nasal Colonization**

**Liliane Maria Fernandes de Oliveira 1,†,‡, Marina Steindorff 2,†, Murthy N. Darisipudi 1, Daniel M. Mrochen 1, Patricia Trübe 1, Barbara M. Bröker <sup>1</sup> , Mark Brönstrup <sup>2</sup> , Werner Tegge 2,\* and Silva Holtfreter 1,\***


**Abstract:** Due to increasing mupirocin resistance, alternatives for *Staphylococcus aureus* nasal decolonization are urgently needed. Adhesion inhibitors are promising new preventive agents that may be less prone to induce resistance, as they do not interfere with the viability of *S. aureus* and therefore exert less selection pressure. We identified promising adhesion inhibitors by screening a library of 4208 compounds for their capacity to inhibit *S. aureus* adhesion to A-549 epithelial cells in vitro in a novel automated, imaging-based assay. The assay quantified DAPI-stained nuclei of the host cell; attached bacteria were stained with an anti-teichoic acid antibody. The most promising candidate, aurintricarboxylic acid (ATA), was evaluated in a novel persistent *S. aureus* nasal colonization model using a mouse-adapted *S. aureus* strain. Colonized mice were treated intranasally over 7 days with ATA using a wide dose range (0.5–10%). Mupirocin completely eliminated the bacteria from the nose within three days of treatment. In contrast, even high concentrations of ATA failed to eradicate the bacteria. To conclude, our imaging-based assay and the persistent colonization model provide excellent tools to identify and validate new drug candidates against *S. aureus* nasal colonization. However, our first tested candidate ATA failed to induce *S. aureus* decolonization.

**Keywords:** *Staphylococcus aureus*; colonization; mouse; JSNZ; aurintricarboxylic acid; ATA; adhesion inhibitor; mupirocin; nose

#### **1. Introduction**

Nasal colonization with *Staphylococcus aureus* is a major risk factor for invasive staphylococcal infections [1,2]. In particular, infections caused by methicillin-resistant *S. aureus* (MRSA) have limited treatment options and are associated with higher morbidity and mortality [3]. To prevent endogenous infection as well as transmission within the hospital, newly admitted patients are routinely screened for MRSA colonization and decolonized using the antibiotic mupirocin [4]. However, increasing bacterial resistance to mupirocin with a prevalence exceeding 13% for MRSA [5] and restrictions for its use have created a need for alternatives [6,7]. In the past, several alternative interventions for the clearance of *S. aureus* nasal carriage have been explored, including antibiotics, such as neomycin [8], polysporin [9], and bacitracin [10], bacteriocins such as lysostaphin [11], as well as fatty acid derivatives (lauric acid monoesters) [12], and cationic synthetic polymers [13]. However,

**Citation:** Fernandes de Oliveira, L.M.; Steindorff, M.; Darisipudi, M.N.; Mrochen, D.M.; Trübe, P.; Bröker, B.M.; Brönstrup, M.; Tegge, W.; Holtfreter, S. Discovery of *Staphylococcus aureus* Adhesion Inhibitors by Automated Imaging and Their Characterization in a Mouse Model of Persistent Nasal Colonization. *Microorganisms* **2021**, *9*, 631. https://doi.org/10.3390/ microorganisms9030631

Academic Editor: Rajan P. Adhikari

Received: 1 February 2021 Accepted: 14 March 2021 Published: 18 March 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

most of these candidates have failed to completely eradicate *S. aureus* nasal colonization. Hence, new approaches to combat *S. aureus* colonization are urgently required.

The first step towards an effective colonization of *S. aureus* in the nares and on other sites of the human body is the attachment of the bacteria to human epithelial cells. This process is facilitated by multifunctional and redundant adhesins on the staphylococcal surface that bind to host cell molecules. An up-to-now underexplored possibility to prevent or eliminate *S. aureus* colonization is to specifically interfere with this bacterial adhesion. Adhesion inhibitors that do not interfere with the viability of the bacteria are promising new preventive agents, because they exert less selective pressure and therefore do not foster the development of resistances [14,15]. Initial studies have been carried out in this context, but so far, they did not lead to clinically useful compounds [16].

The identification of adhesion inhibitors from large compound libraries requires high throughput screening approaches. In former investigations on the adhesion of *S. aureus* to epithelial cells, adherent bacteria were quantified by visual inspection with a microscope and counting [17] or by employing radioactively-labeled bacteria [18]. Alternatively, the adherent bacteria were quantified by determining colony-forming unit (CFU) values after their detachment from eukaryotic cells [19]. These approaches are impractical for the screening of hundreds and thousands of compounds, and they do not detect morphologic effects of the test compounds on the eukaryotic binding partner. Recently, a procedure was reported that is amenable to higher throughput. It utilizes an ELISA to measure binding of *S. aureus* to the major eukaryotic interaction partners fibronectin, keratin, and fibrinogen [20]. However, influences of the compounds on the phenotype of the eukaryotic cells cannot be detected in this screening procedure either.

The in vivo validation of candidate adhesion inhibitors requires a robust and sustained *S. aureus* colonization model. We have previously established a persistent murine nasal colonization model using the mouse-adapted *S. aureus* strain JSNZ [21,22]. Due to its long-term adaptation to the murine host, this strain is capable of inducing persistent colonization of the murine nose and gastrointestinal tract without the need for prior antibiotic treatment [23]. This model enables researchers for the first time to study host– pathogen interaction during persistent colonization in the mouse and also to evaluate decolonization drugs.

Here we report on the development and application of a high throughput microtiter plate-based phenotypic in vitro assay that quantifies the adhesion of *S. aureus* to human epithelial cells. The procedure utilizes fluorescence labeling of eukaryotic cell nuclei and bacteria after their adhesion, followed by detection with an automated microscope with image analysis, in combination with a pipetting robot for the distribution of substance libraries and for liquid handling. We performed a medium throughput screen of more than 4000 compounds, and subsequently characterized aurintricarboxylic acid (ATA) that was identified as the most potent adhesion inhibitor in this assay. The ability of ATA to eradicate nasal *S. aureus* colonization was assessed in our persistent *S. aureus* colonization model. While the standard-of-care antibiotic mupirocin completely eliminated *S. aureus* colonization within three days of treatment, ATA did not show an anti-bacterial effect in vivo. Nevertheless, our novel microscopy-based screening approach and the mouseadapted strain JSNZ are powerful tools to identify and validate new drug candidates against *S. aureus* nasal colonization.

#### **2. Materials and Methods**

#### *2.1. Epithelial Cells*

For the screening procedure, the human epithelial lung cell line A-549 (ACC 107) was obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ, Braunschweig, Germany). For the verification of the activity of hit substances, Human Nasal Epithelial Primary Cells (HNEPC) from Provitro GmbH (Berlin, Germany) were used. Both cell lines were cultivated under conditions recommended by their respective

depositors. Cell culture reagents came from Invitrogen (Carlsbad, CA, USA) and Provitro GmbH (Berlin, Germany).

#### *2.2. S. aureus Strains*

For the screening procedures, the *S. aureus* strain N315 (ST5-CC5-MRSA) was used [24], kindly provided by Prof. Dr. K. Becker, University of Greifswald, Germany. For the development of the assay conditions *S. aureus* SA113 [25] (ST8-CC8-Methicillin-sensitive *S. aureus* (MSSA)) and its adhesion-deficient deletion mutant Delta tagO were kindly provided by Prof. Dr. A. Peschel, University of Tübingen, Germany [26]. For hit validation, the wild type *S. aureus* strains 50,307,270 (rifampicin resistant), 50,128,509 (MRSA), and 50,046,981 (MSSA) were kindly provided by the Institute for Microbiology, Immunology and Hospital Hygiene from the city hospital of Braunschweig, Germany. For the in vivo experiments, the mouse-adapted *S. aureus* strain JSNZ (ST88-CC88-MSSA) was employed [21].

#### *2.3. Compounds*

Six substance collections were investigated (Table 1). Aurintricarboxylic acid (ATA) was initially part of the LOPAC collection. For further testing, it was obtained from Sigma-Aldrich/Merck (Darmstadt, Germany) as the free acid.



<sup>1</sup> Academic collection of the Helmholtz Centre for Infection Research (HZI), sourced from in-house myxobacterial research [27]. <sup>2</sup> Sigma-Aldrich/Merck (Darmstadt, Germany); LOPAC = abbreviation for 'Library of Pharmacologically Active Compounds'. <sup>3</sup> Academic collection of the HZI, sourced from multiple medicinal chemistry groups; VAR stands for 'various sources'. <sup>4</sup> X = mixture of all proteinogenic amino acids, 1 and 2 = defined amino acids (all proteinogenic amino acids in D-configuration), Cys excluded; DKP = diketopiperazine. <sup>5</sup> X = mixture of all proteinogenic amino acids, 1 and 2 = defined amino acids (all proteinogenic amino acids in L-configuration), Cys excluded; DKP = diketopiperazine; 4,5 prepared in the peptide synthesis facility of the HZI. <sup>6</sup> 1 and 2 = all proteinogenic amino acids in D-configuration, Cys excluded [28].

#### *2.4. Mice*

Female C57BL/6NRj mice with Specific and Opportunistic Pathogen Free status (SOPF, *S. aureus*-free, 9 weeks old) were purchased from Janvier Labs (Saint-Berthevin, France). Females were selected because they are less prone to develop genital abscesses upon JSNZ colonization than males [21]. Moreover, males tend to be more aggressive and to present fight wounds that might get infected with *S. aureus*. After delivery, the animals were acclimatized for 7 days before starting the experiments. The animals were kept in individually ventilated cages (IVC, 4 animals/cage) under SOPF conditions with litter material. Food and water (acidified with HCl) were provided ad libitum.

#### *2.5. In Vitro Adhesion Assay*

A-549 cells were resuspended in RPMI 1640 containing 10% FCS and seeded at a density of 1 × <sup>10</sup><sup>4</sup> cells per well in 100 <sup>μ</sup>L medium into 96 well, black, optical bottom microtiter plates (sterile, cell culture-treated; order no. 165,305, Nalgene Nunc, Rochester, NY, USA). After an incubation period of 4 to 5 days under cell culture conditions at 37 ◦C, when the cells had formed a uniform confluent layer at the bottom of the wells, the incubation medium was removed and replaced by 75 μL infection medium (RPMI 1640 containing 1% FCS and 20 mM Hepes, pH 7.4). Test compounds were added with a pipetting robot equipped with a pin tool at a final concentration of 20 μM (pipetting robot: Evolution P3, PerkinElmer, Waltham, MA, USA; pin tool: FP3CB, 96 floating tube pins, 0.787 mm diameter, length: 33 mm, transfer volume 80 nL, V&P Scientific, Inc., San Diego, CA, USA). In each microtiter plate, four wells each of the following controls were included: Cell culture medium containing DMSO in concentrations corresponding to the substance testing and as positive control, 50 μg/mL polyinosinic acid (Sigma-Aldrich, Darmstadt, Germany), which has been shown before to reduce the adhesion of *S. aureus* to epithelial cells by approximately 50% at this concentration [29].

25 μL of *S. aureus* N315 suspension that had been grown overnight in Brain Heart Infusion (BHI) medium (Sigma-Aldrich, Darmstadt, Germany), washed twice with phosphatebuffered saline (PBS), and re-suspended in infection medium to an optical density (OD) OD600 = 1.0 (1 × <sup>10</sup><sup>8</sup> CFU/mL), was added to each well, resulting in an assay concentration of OD600 = 0.25 (2.5 × <sup>10</sup><sup>6</sup> CFU/well). After an incubation period of 1 h at room temperature (RT), non-adherent bacteria were removed by carefully washing the cell layer three times with PBS. The A-549 cells with adherent bacteria were fixed with 4% paraformaldehyde in PBS for 20 min at RT and afterwards washed twice with PBS. For the detection of adherent bacteria 50 μL/well of a primary rabbit antibody against *S. aureus* lipoteichoic acid (Acris Antibodies GmbH, Herford, Germany) diluted 1:5000 in PBS with 1% bovine serum albumin (BSA) was added. After 35 min at RT, 50 μL of 4 ,6 diamidino-2-phenylindoldihydrochloride (DAPI-solution, Sigma-Aldrich, Darmstadt, Germany) for the detection of A-549 cell nuclei was added at a final concentration of 1 μg/mL. For some preliminary experiments in the course of evaluation, the cytoplasm was stained with CellTrackerTM Red CMTPX (Molecular Probes®, Invitrogen, Carlsbad, CA, USA) by incubating with a final concentration of 5 μM together with DAPI. After 10 min at RT and three washing steps with PBS, 50 μL/well of the secondary mouse antibody anti-rabbit Alexa Fluor® 488 (Invitrogen) was added at a dilution of 1:1000 in PBS/1% BSA. After an incubation time of 45 min and washing with PBS, bacteria and A-549 cells were analyzed with the automated microscope ImageXpress Micro (IXM, Molecular Devices, Sunnyvale, CA, USA) and the dedicated software MetaXpress. The initial screening of the substance collections was carried out in duplicate in two independent experiments. In the reevaluation of the initial hits, different concentrations of active compounds were used (5, 10, 20, 50, 100 μM, for ATA also lower concentrations).

For the determination of the effect of ATA on precolonized cells, the adhesion assay was modified: After the initial 1 h incubation of the A-549 cells with bacteria without addition of compounds, the wells were washed carefully three times with PBS. Fresh infection medium and ATA were added to the cells at concentrations of 0.95 and 2.2 μg/mL. After further incubations for 1, 2, 3, 4, and 5 h, the cells were washed three times with PBS, followed by the microscope-based quantification procedure. For each time point controls without ATA were carried out, providing the reference values.

#### *2.6. Quantification of Epithelial Cells and Bacteria*

Each plate was imaged with the automated microscope. For each well, nine images from different sites with a size of 0.4 × 0.4 mm were acquired. A 20× objective and the fluorescence filters "DAPI" (377 and 447 nm for excitation and emission, respectively) for the detection of A-549 nuclei and "FITC" (475 and 536 nm for excitation and emission, respectively) for the quantification of Alexa Fluor® 488 labeled bacteria were used. For some preliminary experiments in the course of evaluation, stained cytoplasm was visualized with the fluorescence filter "Texas Red" (560 and 624 nm for excitation and emission, respectively). Each image was analyzed with the MetaXpress software modul "Transfluor" and the mode "Vesicle area per cell". With this mode, the bacterial area per DAPI stained cell nucleus was quantified. The average was calculated from the nine different images per well. Performance of the assay was evaluated in 92 wells without added compounds; 4 wells contained 50 μg/mL polyinosinic acid as positive control. In addition, the signal from the adherent bacteria (Alexa Fluor® 488 fluorescence; excitation/emission at 485 nm/535 nm) was determined with a

fluorescence microtiter plate reader (Fusion Universal Microplate Analyser, PerkinElmer, Waltham, MA, USA).

#### *2.7. Cytotoxicity Assay*

The cytotoxicity of our hit compounds on A-549 cells was quantified using a 3-(4,5 dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay following the procedure by Mosmann [30], modified by Sasse [31]. Briefly, sub-confluent A-549 cells were washed with PBS without Ca2+ and Mg2+, trypsinized, and resuspended in DMEM containing 10% FCS. 60 μL of serial dilutions of the test compounds were added to 120 μL aliquots of a cell suspension (5000 cells) in 96 well microtiter plates in duplicate. Blank and solvent controls were incubated under identical conditions. After 24 h, 20 μL MTT in PBS was added to a final concentration of 0.5 mg/mL. After 2 h, the precipitate of formazan crystals was centrifuged and the supernatant was discarded. The precipitate was washed with 100 μL PBS and dissolved in 100 μL 2-propanol containing 0.4% hydrochloric acid. The microplates were gently shaken for 20 min to ensure a complete dissolution of the formazan, and finally the absorption was measured at 595 nm using an ELISA plate reader. The percentage of viable cells was calculated and the mean was determined with respect to the controls with medium only.

#### *2.8. Investigation of the Antimicrobial Activity*

Aliquots of 120 μL of an overnight culture of *S. aureus* N315 in BHI medium were washed, adjusted to OD600 = 0.015, corresponding to approximately 5 × 105 CFU/mL, and added to 60 μL of a serial dilution of the test compounds in BHI. After an incubation time of 18 h at 37 ◦C without shaking under moist conditions, the OD600 was measured with a microtiter plate reader (Fusion Universal Microplate Analyser, PerkinElmer, Waltham, MA, USA). The lowest concentration that completely suppressed growth defined the MIC values.

#### *2.9. Preparation of the S. aureus Inoculum for In Vivo Experiments*

*S. aureus* JSNZ strain was grown over night in BHI medium at 37 ◦C and 200 rpm. Thereafter, the culture was diluted to OD595 = 0.05 in BHI medium and cultivated at 37 ◦C and 200 rpm until the mid-logarithmic phase (OD595 = 2.0–2.5). Cells were harvested by centrifugation for 10 min at 8000× *g*, resuspended in BHI containing 14% sterile glycerine, and frozen at −80 ◦C. Before intranasal inoculation, bacteria stocks were thawed on ice, washed once with 40 mL PBS, and reconstituted in PBS to the desired concentration (1 × <sup>10</sup><sup>10</sup> CFU/mL) based on the optical density of the suspension (OD595 = 1; equals 2.6 × <sup>10</sup><sup>8</sup> CFU/mL). The actual bacterial dose was determined by plating serial dilutions of the inoculum on LB agar plates right after intranasal colonization. Plates were incubated over night at 37 ◦C and CFU were counted the following day.

#### *2.10. Preparation of Substances for Intranasal Application*

For initial experiments, ATA was dissolved in three different carrier substances for intranasal application: Poloxamer 407, Softisan 649/Vaseline 9:1, and PBS. Poloxamer 407 powder (Sigma-Aldrich, Darmstadt, Germany) was dissolved in sterile distilled water to the desired concentration and homogenized over night by gently shaking at 4 ◦C. This substance was stored at 4–8 ◦C and maintained on ice before intranasal inoculation. Softisan 649/Vaseline 9:1 (*w*/*w*) (Fagron GmbH & Co. KG, Barsbüttel, Germany), hereafter referred to as S/V, was warmed up to 65–80 ◦C for 10–30 min to obtain a liquefied solution suitable for pipetting. Since mupirocin 2% (Turixin®, GlaxoSmithKline GmbH & Co, München, Germany) was also formulated with S/V, this drug was also liquefied by warming to 80 ◦C for 15 min to enable pipetting of small volumes. ATA (Sigma-Aldrich, Darmstadt, Germany) was dissolved in sterile distilled water to 10–20%, thereafter diluted to the required concentration with Poloxamer (20% *w*/*v*) or with liquefied S/V.

#### *2.11. Treatment of Colonized Mice with Drug Carriers, ATA and Mupirocin*

To exclude a direct effect of the drug carrier itself on the bacterial load, we determined the impact of the three drug carriers PBS, 20% Poloxamer 407, and Softisan 649/Vaseline 9:1 (*w*/*w*) in a persistent *S. aureus* nasal colonization model. Mice were colonized intranasally with 0.7–1.0 × <sup>10</sup><sup>8</sup> CFU *S. aureus* JSNZ (5 <sup>μ</sup>L per nostril) under mild isoflurane anesthesia. Starting on day 3 after colonization, mice were treated once daily for 7 consecutive days with 10 μl of the respective carriers or left untreated. All substances were applied into the nasal cavity using a Hamilton© syringe (100 μL Microliter Syringe, 22s-gauge, Hamilton, Bonaduz, Switzerland) with a shortened and blunted needle. The blunted needle was inserted a few millimeters into the anterior nares to ensure application of the substances within the nasal cavity. Gastrointestinal colonization was examined via stool samples that were collected from individual mice in septic cages on day 0, 3, 6, and 10 after colonization. Mice were weighed and visually inspected for any symptoms of infection on a daily basis. On day 6 and 10 after colonization, mice were sacrificed by isoflurane overdose. The nose including the nasal cavities as well as the cecum were collected for evaluation of the bacterial load.

To determine the anti-adhesive capacity of ATA, mice were colonized with JSNZ for three days as described above and afterwards treated once per day for consecutive 7 days with ATA or left untreated. Five microliters of ATA were slowly applied in each nostril (10 μL in total) using the blunted Hamilton© syringe. The compound was applied at different concentrations (0.5%, 2%, 5%, or 10%) in two different carriers. ATA-Poloxamer was cooled on ice while ATA-S/V was warmed to 65–80 ◦C to turn liquid before intranasal application. The reference substance mupirocin was liquefied at 65–80 ◦C and applied in the clinically used dose of 2.0% (5 μL/nostril) following the same procedure. Both ATA and mupirocin were not affected by heating in this temperature range as verified by LC-MS controls and mupirocin sensitivity testing, respectively (data not shown) [8]. Animals were sacrificed by isoflurane overdose after 0, 3, or 7 days of treatment, and nose and cecum were obtained as detailed above.

#### *2.12. Determination of the Bacterial Load*

Stool samples were adjusted to 0.2 g/mL with sterile PBS followed by homogenization for 20 min at 1400 rpm and 4 ◦C using a Thermo Mixer C shaker (Eppendorf, Hamburg, Germany). The nose and cecum were weighed, transferred to autoclaved homogenizer tubes containing zirconium oxide beads (diameter: 1.4/2.8 mm, Precellys, France), and filled up with 1 mL sterile PBS. Cecum samples were homogenized at 6000 rpm for 2 × 20 s with a 15 s interval. Noses were homogenized at 6500 rpm for 2 × 30 s with a 5 min interval. Homogenized samples were serially diluted; 10 μL of each dilution were plated out in triplicate on *S. aureus* Chromagar plates (CHROmagar, France) and enumerated the next day.

#### *2.13. Ethics Statement*

Animal experiments received ethical approval from the responsible State authorities (Landesamt für Landwirtschaft, Lebensmittelsicherheit und Fischerei Mecklenburg-Vorpommern, 7221.3-1-018/19). The experiments were performed in accordance with the German Animal Welfare Act (Deutsches Tierschutzgesetz), the EU Directive 2010/63/EU for animal experiments and the Federation of Laboratory Animal Science Associations (FELASA). All animal experiments comply with the ARRIVE guidelines.

#### *2.14. Statistics*

Data analysis was performed using the GraphPadPrism6 package (GraphPad Software, Inc., La Jolla, California, USA). Group-wise comparisons were conducted using the Mann-Whitney test or Welch's unpaired *t*-test, as indicated in the particular graph. Paired samples (i.e., disease activity scores) were compared using the Friedman test and Dunn's multiple comparison test for post hoc analyses.

#### **3. Results**

#### *3.1. Development of a Screening Method for the Identification of S. aureus Adhesion Inhibitors*

Our first aim was the development of an automated microscopy-based method that allows the screening of several thousand substances for compounds that inhibit the adhesion of *S. aureus* to human lung epithelial cells (A-549) [17,32–34]. Since the bacteria were found to adhere to the plastic surface of the microtiter plates (not shown), it was important to grow the A-549 cells to a dense confluent layer before performing the adhesion assay. Initial experiments showed a linear correlation between the amount of bacteria used and the bacterial adhesion to A-549 cells (Figure S1). For the subsequent screening campaigns we found an intermediate amount (OD600 = 0.25, corresponding to 2.5 × 107 CFU/mL) to be optimal for the performance of the assay. An incubation time of one hour at RT was found most suitable, since longer incubation times or higher temperatures led to a bias due to bacterial proliferation and/or compounds that have an influence on bacterial growth rather than on adhesion. For the reliable detection and quantification of the bacteria in the automated microscope, fluorescence labeling was employed, and several methods were evaluated, including pre-incubation of the bacteria with fluorescein isothiocyanate (FITC) and Syto 9. The most suitable approach was paraformaldehyde fixation at the end of the incubation period, followed by the staining of adherent bacteria with a commercial primary antibody against *S. aureus* lipoteichoic acid and an Alexa Fluor 488-conjugated secondary antibody, together with DAPI staining of the eukaryotic cell nuclei.

During assay development, we investigated the adhesion of *S. aureus* N315 to A-549 cells in relation to the growth phase. *S. aureus* was cultivated for 2 h, 4 h, 6 h, 8 h, 19 h, and 24 h in BHI, harvested, washed, and resuspended in infection medium. The adhesion of overnight grown bacteria was comparable to the adhesion of mid-log phase bacteria (Figure S2). For technical reasons we preferred to use overnight cultures in our approach.

The adherence was measured with an automated microscope by a quantitative detection of the 'vesicle area per cell', the "vesicles" representing the Alexa Fluor-labeled bacteria and the "cells" the DAPI-stained nuclei of the A-549 lung epithelia. In parallel, adherent bacteria were quantified by scanning the plates with a fluorescence microtiter plate reader. The microscopic approach led to more reliable and reproducible values than the data obtained with the microtiter plate reader. The reliability of both assays was assessed by performing the adhesion test in 92 wells without addition of compounds, including controls (Figure S3). For the microscopy-based test, the Z' factor was determined to be 0.54, which was considered to be of sufficient quality for the screening procedure [35], whereas the Z' factor for the microtiter plate-based test was only 0.12 and considered insufficient.

The functionality of the assay in terms of the detection of influences on adhesion was validated with the *S. aureus* wild type strain SA113 and its isogenic mutant SA113 Delta tagO, whose efficiency of adhesion was reduced by approximately 50% as shown before [34]. Different ratios of bacteria to A-549 cells showed the expected differences between the strains (Figure S1).

The cell nuclei and the adherent bacteria were reliably identified by automated microscopy and image analysis. As illustrated in Figure 1, there was a good match between the microscopic pictures of the bacteria and the A-549 cells after adhesion, fixation, and fluorescence staining (upper pictures) and the results from the automated image analysis and the labeling of the recognized structures (lower pictures). By staining the cytoplasm of the A-549 cells in addition to the cell nuclei and the bacteria, we verified during the assay development that the bacteria did not migrate from their initial attachment site on the A-549 cells to sites on the well surface where the eukaryotic cells have been detached during the washing procedures (Figure S4). To conclude, the established automated microscope-based approach enabled the quantitative investigation of adhesion processes in microtiter plates and additionally revealed morphological changes that are induced to the cells upon exposure to the tested substances.

**Figure 1.** Identification of adherent bacteria by automated microscopy and image analysis. Adhesion of wild type *Staphylococcus aureus* SA113 (**left**) and *S. aureus* SA113 Delta tagO (**right**) to A-549 cells is depicted. Cell nuclei were stained with DAPI (**upper panel**, blue and **lower panel**, green) and bacteria were stained with a primary antibody against teichoic acid and an Alexa 488-labeled secondary antibody (**upper panel**, green and **lower panel**, red). Upper panel: Images acquired with the automated microscope after washing and staining; lower panel: Bacterial and epithelial cell areas as identified by the imaging software MetaXpress with the module "Transfluor" and the mode "Vesicle area per cell", based on the microscopic images shown in the upper panel. A bacterial density of OD600 = 0.1 has been used (for quantification please refer to Figure S1).

#### *3.2. Screening of Compound Libraries for S. aureus Adhesion Inhibitors*

After the optimization and validation of the assay conditions, bacterial adhesion was investigated in a screening approach with 4208 compounds (Table 1) at 20 μM concentration in duplicate. The compound collection consisted of six different parts. Our in-house collection of secondary metabolites from myxobacteria (part 1) is a unique source of highly bio-active compounds. Many of those compounds have shown antibacterial, antiherbal, cytotoxic, and other properties and are the focus of ongoing investigations by us and many others [27,29]. The LOPAC collection of pharmacologically active compounds (part 2) is a broad collection of a large number of drugs and drug-like molecules, most of which have already been studied in preclinical and clinical investigations. Hits that would be derived from the collection have the advantage that an approval as a drug can be aided by prior data that is already available, which may speed up the process considerably and reduce cost ("drug repurposing"). Our front-runner ATA was part of the LOPAC collection. The VAR collection (part 3) is also unique at our institute and provides a variety of unusual structures that are not found in most commercial substance collections and extended the "structure space" of our screening campaign considerably. The linear and cyclic peptides (part 4–6) are consisting of sublibraries with defined and randomized positions. Such libraries have been used successfully in other screening approaches by employing an iterative stepwise procedure to enhance biological activities. We considered this combination of unique and drug-like structures a solid basis for our project.

The adhesion inhibitor polyinosinic acid was used at 50 μg/mL as a positive control in each microtiter plate [36]. In the primary screen, 62 compounds were found to be active with at least 30% reduction of adhesion (Supplementary Tables S1 and S2). Substances that caused substantial loss of eukaryotic cells during the incubation were excluded from further testing. An advantage of using the automated microscope was the possibility to reinvestigate wells individually by visual inspection of the pictures that were recorded. Since the ratio of bacteria to cell nuclei was determined, the approach was, contrary to the microtiter plate readings, insensitive to partial losses of the epithelial cells during the washing procedures. The antimicrobial effect of the initial hits was evaluated in a broth dilution assay (MIC determinations). Substances with MIC values below 100 μM were excluded from further testing. After the re-evaluation process three compounds were identified that reliably and selectively reduced bacterial adhesion without detaching the eukaryotic cells and with MIC values > 100 μM (Table S3 and Figure S5; for structures see Figure 2B and Figure S6).

#### *3.3. ATA Inhibited S. aureus Adhesion to Epithelial Cells In Vitro*

The most active substance found in the screen was polyaromatic aurintricarboxylic acid (ATA, Figure 2B), which reproducibly reduced bacterial adhesion with a half-maximal inhibitory concentration (IC50) of 0.95 μg/mL. The maximum effect size was a reduction of adhesion to ca. 20% of the original value; this was reached at a compound concentration of approximately 2.2 μg/mL (Figure 2A).

Over the course of the adhesion experiment, ATA did not cause any apparent morphological effect on the A-549 cells. In addition, bacterial growth was not impaired by ATA up to a concentration of 42 μg/mL (100 μM, data not shown). In order to assess whether ATA exerts cytotoxic effects, an MTT test with A-549 cells was carried out. After 24 h of incubation, ATA caused less than half-maximum inhibition of metabolic activity of the A-549 cells up to the highest test concentration of 370 μg/mL (data not shown). Toxicities for ATA in mice have been reported before. A dose of 100 mg/kg/day per day i.p. was found to be lethal, whereas 30 mg/kg/day was not [37].

To analyze the activity of ATA towards different *S. aureus* strains, three clinical isolates of *S. aureus* were used, strain 50,307,270 (rifampicin resistant), strain 50,128,509 (MRSA), and strain 50,046,981 (MSSA). ATA could inhibit the adhesion of all strains with a potency that was equal to or higher than that found for N315 (Figure 2C).

Next, the effect of ATA on HNEPC was investigated. The results obtained with the immortalized epithelial cell line A-549 could be confirmed: The adhesion of *S. aureus* N315 to the primary cells was reduced with an IC50 of approximately 1 μg/mL (Figure 2D). Again, ATA did not completely block adhesion, but reduced it to approximately 40% of the original level.

In the primary screening assay, the compounds were already present when bacteria were added to the host cells. Thus, the assay mimics a preventive setting. In order to investigate whether ATA might also work in a therapeutic setting, its effect on already adherent bacteria was investigated next. For this purpose, A-549 cells were precolonized with *S. aureus* N315 for one hour, followed by the removal of non-adherent and planktonic bacteria by washing. Further incubations from one to five hours with 0.95 μg/mL and 2.2 μg/mL ATA (corresponding to the IC50 and maximal effective concentrations, respectively) were carried out. In comparison to the controls without an active compound, ATA was able to reduce the number of bacteria from precolonized A-549 cells in a dose- and time-dependent manner (Figure 2E). This demonstrates that ATA can disrupt an already existing bacterial adhesion, which suggests that ATA might be used in both preventive and therapeutic settings.

**Figure 2.** Aurintricarboxylic acid (ATA) inhibited *S. aureus* adhesion to A-549 cells and to human primary epithelial cells in a dose-dependent manner and reduced precolonization. (**A**) Adhesion of *S. aureus* N315 to A-549 cells was determined in the presence of different concentrations of ATA using the microscopic adhesion assay; Mean and standard deviation for three replicates are depicted. (**B**) Structure of ATA in its monomeric form. (**C**) ATA reduced the adhesion of *S. aureus* clinical isolates to A-549 cells. Mean and standard deviation for two replicates are depicted. \* *p* < 0.4, \*\* *p* < 0.003, ns = not significant, as compared to control according to Welch's unpaired *t*-test. (**D**) Adhesion of *S. aureus* to primary epithelial cells from human nares (HNEPC) in presence of ATA. Mean and standard deviation for three replicates are depicted. (**E**) Application of ATA to A-549 cells that were pre-colonized with *S. aureus* N315 resulted in the time-dependent detachment of the bacteria; incubation times are indicated. The values for each time point are related to controls without ATA. Mean and standard deviation for three replicates are depicted. \* *p* < 0.014, \*\* *p* < 0.01, ns = not significant, as compared to control according to Welch's unpaired *t*-test.

#### *3.4. Drug Carriers Did Not Interfere with S. aureus Colonization*

A suitable drug carrier has to distribute the active compound throughout the nasal cavity to facilitate *S. aureus* elimination [38], without affecting the bacterial load itself. In a pilot experiment, we compared the intranasal distribution of three drug carriers, Poloxamer 407, S/V, and PBS. We mixed them with Evans blue to visually inspect their distribution in the nasal cavity. Poloxamer 407 and S/V were chosen because of their very different viscosity and physical properties, which could influence their distribution in the nasal compartments. Whereas S/V is highly viscous at room temperature and needs to be warmed up to become liquid and pipettable, Poloxamer 407 is liquid at low temperature and turns into a gel at 37 ◦C. All substances were applied into the nasal cavity by inserting a Hamilton© syringe with a shortened and blunted needle a few millimeters into the anterior nares. All three substances spread equally well throughout the entire nasal cavity 30 min after application. They were found in the ventral region close to the nares, along all nasal turbinates, and also in the most dorsal regions, such as the nasopharynx (data not shown).

Next we studied the influence of the carrier substances on *S. aureus* colonization in vivo using a new mouse model of *S. aureus* nasal colonization/decolonization established by our group [21,22]. Mice were intranasally inoculated with the mouse-adapted *S. aureus* JSNZ strain to induce persistent colonization. Mice were colonized with JSNZ for three days and subsequently treated daily with Poloxamer, S/V or PBS without the active compound for 3 to 7 days. As expected, *S. aureus* JSNZ persistently colonized the nasopharynx and gastrointestinal tract of mice throughout the experiment (Figure 3). The tested drug carriers did not impact on *S. aureus* colonization, as reflected by constant bacterial loads in the nose, cecum, and feces. This advantageous result is in strong contrast to our preliminary findings in the cotton rat model (see below). A complete elimination of *S. aureus* in the nose and cecum was only observed in a single mouse treated for 3 days with S/V. These experiments demonstrated that all tested carrier substances are suitable for the delivery of candidate drugs in our *S. aureus* nasal colonization/decolonization model.

#### *3.5. ATA Failed to Induce S. aureus Decolonization, While Mupirocin Was Highly Effective*

Finally, we investigated whether the hit compound ATA reduces *S. aureus* nasal burden. We used Poloxamer and S/V as drug carriers. After three days of bacterial colonization, mice were treated for 7 consecutive days with ATA ointment in two different doses (0.5% and 2%). Its effect on the *S. aureus* load was compared with the human therapeutic agent mupirocin. As expected, the antibiotic mupirocin eradicated *S. aureus* from the nasal region even in a short treatment regime of 3 days (Figure 4A). Although applied only to the nasal cavity, mupirocin also eliminated *S. aureus* from the gastrointestinal tract of most mice within 7 days of treatment (Figure 4B,C). In contrast, ATA did not reduce the *S. aureus* burden in the nose, cecum or feces. This was true at concentrations of 0.5% (2.5 mg/kg body weight) or 2% (10 mg/kg body weight) and in combination with Poloxamer or S/V. Even very high ATA concentrations of up to 10% just reaffirmed the inefficacy of the compound in reducing *S. aureus* colonization. On the contrary, ATA treatment even increased the median nasal burden of *S. aureus* JSNZ (Figure 4). Moreover, mice treated with 10% ATA showed a slight reduction of weight when compared with untreated mice, resulting in a significantly higher disease activity index of this group (*p* = 0.0473 for 10% ATA-S/V; Figure S7). Altogether, the in vivo experiments demonstrate the robustness and suitability of our murine *S. aureus* nasal colonization/decolonization model for testing candidate drugs for *S. aureus* decolonization. ATA, however, failed to induce *S. aureus* decolonization.

**Figure 3.** Drug carriers did not interfere with the bacterial burden in a persistent *S. aureus* colonization model. Female C57BL/6N mice were colonized intranasally with 0.7–1.0 <sup>×</sup> 108 CFU *S. aureus* JSNZ. Starting on day 3 after colonization, mice were treated once daily for 7 consecutive days with 10 μL of the carrier substances PBS, Softisan 649/Vaseline, Poloxamer 407 or left untreated. The *S. aureus* burden was determined in the homogenized nose (**A**), cecum (**B**), and feces (**C**) 0, 3, and 7 days after starting the treatment. The detection limit is indicated by a dashed line; medians are indicated. Data were pooled from two independent experiments with 4 mice/group. Abbreviations: Ctrl—control; PBS—phosphate-buffered saline; Pol.—Poloxamer 407; S/V—Softisan 649/Vaseline 9:1 (*w*/*w*).

**Figure 4.** Mupirocin eradicated *S. aureus* colonization, while ATA was not effective. Female C57BL/6N mice were colonized intranasally with 0.7–1.0 <sup>×</sup> <sup>10</sup><sup>8</sup> CFU *S. aureus* JSNZ. After three days of colonization, mice were treated on a daily basis for 7 days with 10 μL 2% Mupirocin or ATA in different concentrations (0.5%, 2.5%, 5% or 10%) using Softisan 649/Vaseline or Poloxamer 407 as drug carrier. Groups receiving 5% or 10% ATA were only analyzed at day 7. A control group remained untreated. The *S. aureus* bacterial load was determined in the homogenized nose (**A**), cecum (**B**), and feces (**C**) 0, 3, and 7 days after starting the treatment. The detection limit is indicated by a dashed line; medians are indicated. Data were pooled from two independent experiments with 4 mice/group. Statistics: Groups were compared with Ctrl using Mann-Whitney test, \* *p* < 0.05, \*\* *p* < 0.01. Abbreviations: Ctrl—control; Pol.—Poloxamer 407; S/V—Softisan 649/Vaseline 9:1 (*w/w*).

#### **4. Discussion**

Due to increasing numbers of mupirocin-resistant *S. aureus*, new decolonization approaches are urgently required. In this study, we investigated pathoblockers, which means compounds that reduce the pathogenicity of the bacteria without interfering with their viability. Hence, adhesion inhibitors should exert less selection pressure than antibiotics and reduce the development of resistances. In accordance with this goal, we have sorted out compounds that showed antimicrobial effects (MIC values) below 100 μM in our further evaluation of the initial hits.

Our first task was the development of a method that allows the screening of several thousand substances for compounds that inhibit the adhesion of *S. aureus* to epithelial cells. For this purpose, an automated microscope was employed that allows the quantitative investigation of adhesion processes in microtiter plates and that can in addition be used to reveal morphological changes that are induced to the cells by the influence of the substances. Studies of the adhesion of bacteria to respiratory epithelial cells are frequently carried out with the human lung epithelial cell line A-549 [17,24–32], which was also employed in the present investigation.

In the human nose *S. aureus* is in a constant process of exponential growth due to ongoing excretion and shedding [39]. Such a situation suggests good chances for a therapeutic application of adhesion inhibitors. We report a medium throughput screening of more than 4.000 compounds for *S. aureus* adhesion inhibitors and the subsequent characterization of ATA, which was identified as the most potent compound in this assay. Our method works with unlabeled bacteria and eukaryotic cells during the adhesion process, which offers the possibility to use unmodified wild type bacterial strains and to capture any influences of the test compounds on the morphology of the eukaryotic cells in the initial screening campaign. This is an advantage over previously published procedures, as it allows the rapid exclusion of compounds that exert their effect mainly through toxic effects on the eukaryotic cells. It is likely that the approach for identifying adhesion inhibitors presented in this study can be applied to other settings, where the adhesion of pathogens to epithelial cells or eukaryotic cells in general is of interest. Bacterial adhesion to epithelial cells is the important first step in many clinically significant conditions, like the infection of the urinary tract with *Escherichia coli* [40], the colonization of the lungs with *Pseudomonas aeruginosa* [41] or the adherence of Streptococci to the pharynx [42]. By employing specific and adequate staining procedures, the approach described in this study can be adopted to the particular setting.

The number of bacteria that may have internalized into the A-549 cells has not been determined in our current protocol. Previous studies reported a low internalization rate with only 15% of A-549 cells being infected and low bacterial numbers per cell after 1.5 h of incubation [43]. Therefore, we do not expect a relevant influence of internalization events on our results.

For ATA several biological functions have been described [44–50], but the finding that ATA potently prevents the adhesion of *S. aureus* to epithelial cells was new. It should be noted that ATA, which is shown in its monomeric form in Figure 2B, has a strong tendency to oligomerize, which results in polyanionic structures [51]. We speculate that the activity of ATA to reduce adhesion of *S. aureus* to epithelial cells might be linked to its acidic functions, because several other polyanionic substances like polyinosinic acid and sulfated polysaccharides are known to inhibit the cellular adhesion of *S. aureus* [35,52–54]. Whereas the polypharmacological properties and the limited chemical stability render a systemic use of ATA challenging, a topical administration for the decolonization of the nose appeared attractive based on the cellular data.

Apart from the anti-adhesive effect reported here, ATA can act as a *S. aureus* pathoblocker by potent inhibition of Stp1 [55]. Stp1 is a Ser/Thr-phosphatase involved in the global regulation of staphylococcal virulence factors, such as alpha-hemolysin and leukocidins, immune-evasion molecules, such as SCIN and CHIPS, as well as capsular polysaccharide synthesis enzymes [56–59]. Hence, Stp1 inhibition switches *S. aureus*from an invasive, virulent

state to a more silent, immune-evasive state. Administration of ATA significantly reduced the severity of staphylococcal infection in a murine abscess formation model [55].

ATA is also well-known as a small molecule inhibitor of nucleases and nucleic acidbinding enzymes [60]. Diverse studies have reported that ATA inhibits the replication of a variety of viruses in vitro by interference with viral polymerases and RNA-binding proteins [61–63]. Moreover, ATA compromised bacterial biofilm formation by limiting protein-nucleic acid interaction [64]. Whether this is also relevant to *S. aureus* biofilms, which are also composed of eDNA, remains to be investigated.

Despite being the top candidate in the in vitro screening approach, ATA was ineffective in promoting *S. aureus* decolonization in our mouse *S. aureus* colonization model, but even increased the bacterial burden when applied intranasally in a higher dose (10% ATA, equals 50 mg/kg body weight/day). We suggest that the anti-adhesive effect of ATA observed in our in vitro assay could have been masked in vivo by its anti-inflammatory activity [65–68]. It is well known that the innate and adaptive immune responses are critical for the clearance or persistence of *S. aureus* nasal colonization [69–71]. ATA, however, can inhibit innate and adaptive immune responses in many ways. The compound interferes with JAK-STAT signaling, thereby inhibiting IFN-G-induced iNOS expression [68]. Moreover, ATA is able to block chemotaxis of dendritic cells and T cells [67], to inhibit complement activation in vitro [65], and to convert naïve CD4+ T cells into FoxP3+ regulatory T cells [66]. Collectively, these mechanisms could reduce the immune response against the colonizing staphylococci and thereby enhance bacterial persistence in the nasal cavity. In consequence, such an anti-inflammatory effect could override the anti-adhesive properties of ATA in the nasal colonization model. ATA's anti-inflammatory effect could also explain other cases where a protective effect shown in vitro could not be reproduced in the living organism. In line with our data, intraperitoneal injection of ATA enhanced disease severity in murine models of vaccinia virus and orbivirus infection [44,72].

The in vivo evaluation of novel decolonization drugs requires a robust and persistent *S. aureus* colonization model with high discriminatory power between treated and untreated animals. In the past, researchers usually colonized laboratory mice with humanadapted clinical *S. aureus* isolates. Due to poor adaptation to the murine environment, human *S. aureus* isolates usually do not consistently colonize mice and are frequently eliminated from the murine nasal cavity within days [12,21,73,74]. Nevertheless, several drug candidates were investigated using this model [8,12]. However, the variable colonization rate and the rapid natural decolonization strongly reduced the signal-to-noise-ratio in these studies [12].

In search for a more robust and reliable animal model, Kokai-Kun et al. established a nasal colonization model in cotton rats [11]. In cotton rats there is consistent and persistent high-level (~5000 CFUs/nose) nasal colonization by *S. aureus* and consequently a high discriminatory power. This model has been used over the past two decades to evaluate antimicrobials, e.g., lysostaphin [11], cationic methacrylate polymers [13], epidermicin [75], and antimicrobial peptides [76]. Cotton rats are highly excitable rodents and hence much more difficult to handle than mice [77]. We carried out initial studies with ATA in the cotton rat model using human-adapted *S. aureus* strains. However, several drug carriers without drug led to a drastic decrease of the nasal bacterial load within just a few days, which made it impossible to discern any additional drug-related effect (data not shown). Hence, it is essential that the drug carrier distributes the active compound throughout the nasal cavity to facilitate *S. aureus* elimination [38], but does not affect the bacterial load itself. In our study, all three tested substances (Poloxamer 407, S/V, and PBS) spread equally well throughout the entire nasal cavity 30 min after application and did not impact on the *S. aureus* density in the nose.

We have recently reported that laboratory and wild mice are natural hosts of *S. aureus*, and that their colonizing strains show features of adaptation to their murine host. The prototype of these mouse-adapted strains, JSNZ, can colonize the murine nose and gastrointestinal tract for several weeks [21,59]. Colonization was induced in all inoculated

animals with average nasal colonization loads even higher than in the cotton rat model (2 × <sup>10</sup><sup>5</sup> CFU/g nose tissue, corresponding to 4 × 104 CFU/nose). Our consistent and reproducible *S. aureus* colonization model now enables researchers to reliably evaluate novel decolonization drugs in the mouse.

The validity of our animal model is underlined by the results of mupirocin treatment, which served as benchmark for *S. aureus* decolonization. After three days of treatment nasal decolonization was 100%, decolonization of the gastrointestinal tract took a few days longer. In studies with mice that were colonized with human *S. aureus* strains or that used the cotton rat model, mupirocin frequently failed to completely decolonize the nose even at the clinically effective concentration of 2% mupirocin [8,12,13,76]. Apart from the animal model, this might also be due to differences in the experimental settings such as different application modes and schemes. Moreover, the employed carrier substances could have an impact on the spatial distribution of mupirocin within the nasal cavity and its release over time [11].

#### **5. Conclusions**

Adhesion inhibitors are a promising, but yet underexplored option to prevent or eliminate *S. aureus* colonization. The microscopy-based screening presented here is a powerful method to identify and test new *S. aureus* adhesion inhibitors. Our adhesion assay works with unlabeled bacteria and eukaryotic cells, which offers the possibility to use unmodified wild type bacterial strains and to capture any influences of the test compounds on the morphology of the eukaryotic cells in the initial screening campaign. Hence, it allows the rapid exclusion of compounds that exert their effect mainly through toxic effects on the eukaryotic cells. Importantly, this approach can be applied to other settings where the adhesion of pathogens to epithelial cells or eukaryotic cells in general is of interest.

The mouse-adapted strain JSNZ induces persistent nasal colonization and therefore provides an excellent tool for testing drug candidates against *S. aureus* nasal colonization in mice. In contrast to previous models using human-adapted *S. aureus* strains in mice, this model has a high discriminatory power between treated and untreated animals. While cotton rats can also be persistently colonized with *S. aureus*, their handling is much more difficult and less molecular tools are available to study the host response in detail. Even though our first candidate ATA failed to decolonize *S. aureus* from the murine nose, we are optimistic that our screening approach and persistent colonization model will provide a powerful tool for identifying effective antibacterial drugs in the future.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/2076-2 607/9/3/631/s1, Figure S1: Adherence of the *S. aureus* SA113 Delta tagO mutant vs. the wild type strain, Figure S2: Adhesion of *S. aureus* N315 to A-549 cells after precultivation of the bacteria in BHI for different time periods Figure S3: Assessment of the performance of the adhesion test, Figure S4: Visualization of the adhesion of *S. aureus* N315 to A-549 cells, Figure S5: Reevaluation of the adhesionreducing effect of the three most active compounds that were identified in the screening campaigns. Figure S6: Structures of the active compounds HZI10676D08 (methyl 7-hydroxy-9-methyl-6-oxo-6Hoxepino [2,3-b]chromene-5-carboxylate) and HZI10687B10 (pseudohypericin). Figure S7: Evaluation of the health status of the *S. aureus*-colonized mice after treatment with ATA or mupirocin.

**Author Contributions:** Conceptualization, M.B., W.T., B.M.B., and S.H.; formal analysis, L.M.F.d.O., and M.S., methodology, L.M.F.d.O., M.S., W.T., and S.H.; investigation, L.M.F.d.O., M.S., W.T., M.N.D., D.M.M., P.T., and S.H.; data curation, L.M.F.d.O., and M.S.; writing—original draft preparation, L.M.F.d.O., M.S., W.T., and S.H.; writing—review and editing, all co-authors; supervision, W.T., M.B., B.M.B., and S.H.; funding acquisition, M.B., W.T., and S.H. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Bundesministerium für Bildung und Forschung (BMBF), grant number 01Kl0710 (SkinStaph), the Deutsche Forschungsgemeinschaft (DFG), grant number FKZ GRK1870/1, and the Bundesministerium für Bildung und Forschung (BMBF), InfectControl 2020, project InVAC, FKZ 03ZZ0806B to D.M.M.

**Acknowledgments:** We thank A. Peschel, University of Tübingen, Germany, K. Becker, University of Greifswald, Germany, and W. Bautsch, Institute for Microbiology, Immunology and Hospital Hygiene, Braunschweig, Germany, for kindly providing *S. aureus* strains. We thank G. Höfle for providing test compounds. Furthermore, we thank Randi Diestel for excellent support concerning the operation of the IXM, Tabea Ellebracht for technical assistance in performing the screening experiments, Raimo Franke for help with statistical evaluation, Kilian Wietschel and Grazyna Domanska for support with animal dissection, Susan Mouchantat for her support regarding the intranasal application of ointments, and Susanne Neumeister, Erika Friebe, Fawaz Al'Sholui, and Sabine Prettin for technical assistance. We thank Carsten Peukert for support regarding the generation of the graphical abstract.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


## *Article* **Impact of the Histidine-Containing Phosphocarrier Protein HPr on Carbon Metabolism and Virulence in** *Staphylococcus aureus*

**Linda Pätzold 1, Anne-Christine Brausch 1, Evelyn-Laura Bielefeld 1, Lisa Zimmer 1, Greg A. Somerville 2, Markus Bischoff 1,\* and Rosmarie Gaupp <sup>1</sup>**


**Abstract:** Carbon catabolite repression (CCR) is a common mechanism pathogenic bacteria use to link central metabolism with virulence factor synthesis. In gram-positive bacteria, catabolite control protein A (CcpA) and the histidine-containing phosphocarrier protein HPr (encoded by *ptsH*) are the predominant mediators of CCR. In addition to modulating CcpA activity, HPr is essential for glucose import via the phosphotransferase system. While the regulatory functions of CcpA in *Staphylococcus aureus* are largely known, little is known about the function of HPr in CCR and infectivity. To address this knowledge gap, *ptsH* mutants were created in *S. aureus* that either lack the open reading frame or harbor a *ptsH* variant carrying a thymidine to guanosine mutation at position 136, and the effects of these mutations on growth and metabolism were assessed. Inactivation of *ptsH* altered bacterial physiology and decreased the ability of *S. aureus* to form a biofilm and cause infections in mice. These data demonstrate that HPr affects central metabolism and virulence in *S. aureus* independent of its influence on CcpA regulation.

**Keywords:** *Staphylococcus aureus*; physiology; metabolism; carbon catabolite repression; CcpA; HPr

#### **1. Introduction**

Carbon catabolite repression (CCR) is a common regulatory mechanism of bacteria to coordinate central metabolism with available carbon source(s) [1]. By modulating transcription of genes encoding proteins involved in the import and catabolism of carbon metabolites, bacterial CCR facilitates the efficient use of available carbon sources [1]. In pathogenic bacteria, regulators of CCR often affect transcription of virulence factors that are important for the exploitation of host-derived nutrient sources [2].

*Staphylococcus aureus* is a gram-positive opportunistic pathogen and a frequent cause of nosocomial infections in which central metabolism and infectivity are linked by numerous regulatory factors, including the catabolite control proteins A (CcpA) and E (CcpE), CodY, Rex, RpiRc, and SrrAB [3]. CcpA, a member of the GalR-LacI repressor family [4], is thought to be the major factor regulating CCR in *S. aureus* by binding catabolite-responsive element (*cre*) sequences of target genes [5]. Depending on the *cre* sequence location in the promotor region, the binding of CcpA results in either activation or repression of transcription [6]. Studies using *Bacillus megaterium* and *Streptococcus pyogenes* demonstrated that the binding affinity of CcpA for *cre* sites is low, but can be increased drastically by complex formation with the histidine-containing phosphocarrier protein (HPr), encoded by *ptsH* [7,8]. Electrophoretic mobility shift assays suggest this is also true in *S. aureus* [6], although CcpA can also bind to *cre* sites in the absence of HPr [9]. Activity of HPr is dependent on at least two phosphorylation sites, namely amino acids histidine 15 (His-15) and serine 46 (Ser-46) [1]. For complex formation with CcpA, HPr must be phosphorylated

**Citation:** Pätzold, L.; Brausch, A.-C.; Bielefeld, E.-L.; Zimmer, L.; Somerville, G.A.; Bischoff, M.; Gaupp, R. Impact of the Histidine-Containing Phosphocarrier Protein HPr on Carbon Metabolism and Virulence in *Staphylococcus aureus*. *Microorganisms* **2021**, *9*, 466. https://doi.org/ 10.3390/microorganisms9030466

Academic Editor: Rajan P. Adhikari

Received: 7 February 2021 Accepted: 19 February 2021 Published: 24 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

on Ser-46 [7]. This ATP-requiring process is catalyzed by the HPr-kinase/phosphorylase (HPrK/P), which is regulated in a dose-dependent manner by the glycolytic intermediate fructose-1,6-bisphosphate (FBP) [10]. For this reason, the amount of Ser-46 phosphorylated HPr (P-Ser-HPr) is closely connected with glycolytic activity of the cell and the uptake of sugars. Sugar uptake in bacteria is predominantly mediated by the phosphotransferase system (PTS), consisting of three main components: HPr, enzyme I (EI), and enzyme II (EII) [11]. In a first step, HPr is phosphorylated at His-15 (P-His-HPr) by E1, using the glycolytic intermediate phosphoenolpyruvate as the phosphate donor. The phosphate group is transferred to the substrate by EII, which translocates and phosphorylates the sugar into the cell at the same time. Activated glucose, namely glucose 6-phosphate, then enters glycolysis [12]; hence, HPr connects glycolytic activity with CCR via its dual role in sugar uptake through the PTS and as an activator of CcpA [13,14].

Numerous genes have been identified to be regulated on the transcriptional level by CcpA in *S. aureus* [15–17]. In addition to genes/operons involved in carbon catabolism, the synthesis of factors associated with biofilm formation and virulence of *S. aureus* are also influenced by CcpA [6,15–21]. Specifically, CcpA promotes transcription of the *ica*operon and *cidA* [19], encoding proteins needed for polysaccharide intercellular adhesion (PIA) synthesis and extracellular DNA release, respectively [22,23]. These observations are consistent with the fact that deletion of *ccpA* abrogates biofilm formation under glucoserich conditions. [19]. Furthermore, inactivation of *ccpA* in *S. aureus* reduces the formation of liver and skin abscesses in mouse models of infection [6,24,25]. Taken together, these observations demonstrate the linkage between CcpA, glucose catabolism, and virulence in *S. aureus*; however, the function of HPr remains largely unknown. Here, we characterize the function of HPr of *S. aureus* in the context of carbon metabolism, growth kinetics, biofilm formation, and in vivo infectivity in different murine infection models.

#### **2. Materials and Methods**

#### *2.1. Bacterial Strains and Plasmids*

The bacterial strains and plasmids used in this study are listed in Table 1. All mutant strains generated for this study were confirmed by sequencing of the affected region, and by assessing gene transcription by quantitative real-time reverse transcriptase PCR (qRT-PCR).


#### **Table 1.** Strains and plasmids used in this study.


**Table 1.** *Cont.*

<sup>1</sup> KanR, kanamicin-resistant; PIA, polyintercellular adhesin; TcR, tetracycline-resistant.

#### *2.2. Bacterial Growth Conditions*

*S. aureus* strains were grown in tryptic soy broth (TSB) containing 0.25% (*w*/*v*) glucose (BD, Heidelberg, Germany) or on TSB plates containing 1.5% agar (TSA). Antibiotics were only used for strain construction and phenotypic selection at the following concentrations: tetracycline, 2.5 μg/mL; erythromycin, 2.5 μg/mL; kanamycin, 15 μg/mL; and chloramphenicol, 10 μg/mL. Bacteria from overnight cultures were diluted in pre-warmed TSB to an optical density at 600 nm (OD600) of 0.05. All bacterial cultures were incubated at 37 ◦C and 225 rpm with a flask-to-medium ratio of 10:1. Samples for determination of the OD600, pH, and metabolites were taken every hour. The growth rate (*μ*) of *S. aureus* strains was calculated by the formula (ln *OD*2–ln *OD*1)/(*t*2–*t*1), with *OD*<sup>1</sup> and *OD*<sup>2</sup> being the OD calculated from the exponential growth phase at time *t*<sup>1</sup> and *t*2, respectively. The generation time of each strain was determined using the formula ln 2/*μ*.

#### *2.3. Mutant Construction*

For the *S. aureus ptsH* deletion mutants, 1.4- and 1.1-kb fragments (nucleotides 1053472- 1054838 and 1055049-1056169 of GenBank accession no. AP009351.1, respectively), containing the flanking regions of the *ptsH* open reading frame (ORF), were amplified by PCR from chromosomal DNA of *S. aureus* strain Newman using primer pairs MBH-94/MBH-112 and MBH-113/MBH-114, respectively (Supplementary Table S1). The PCR products were digested with KpnI/EcoRI and BamHI/XbaI, respectively, and cloned together with the EcoRI/BamHI-digested lox66-*aphAIII*-lox71 resistance cassette obtained from pBT *loxaph* [26] into KpnI/XbaI-digested suicide vector pBT [27] to generate plasmid pBT *ptsH* KO. Plasmid pBT *ptsH* KO was propagated in *E. coli* strain DC10B [28] and subsequently electroporated directly into *S. aureus* strain Newman to obtain strain Newman Δ*ptsHaph*, in which nucleotides 8 to 225 of the 267-bp spanning *ptsH* ORF were replaced by the lox66-*aphAIII*-lox71 cassette by allelic replacement. The deletion of *ptsH* in Newman Δ*ptsH*-*aph* was confirmed by PCR, and the strain was then used as a donor for transducing the lox66-*aphAIII*-lox71 tagged *ptsH* deletion into *S. aureus* strains SA113 and RN4220. Resistance marker-free Δ*ptsH*::lox72 derivatives were constructed by treatment with a Cre recombinase expressed from the temperature-sensitive vector pRAB1 [29], which was subsequently removed from the *aphIII*-cured derivatives by culturing the strains at 42 ◦C.

For the cis-complementation of the Δ*ptsH*::lox72 mutants, a 1-kb fragment (nucleotides 1055996-1056973 of GenBank accession no. AP009351.1) of the C-terminal region of the *ptsI* ORF and the annotated terminator region of the *ptsHI* operon was amplified by primers MBH-427/MBH-428 (Supplementary Table S1), digested with EcoRI/KpnI and cloned into EcoRI/KpnI-digested suicide vector pBT [27] to generate plasmid pBT '*ptsI*. The plasmid was electroporated into *S. aureus* strain RN4220, and a tetracycline-resistant RN4220 derivative that integrated pBT '*ptsI* in its chromosome at the *ptsI* locus was used as donor to phage-transduce the *tet*(L)-tagged *ptsI* allele into Nm Δ*ptsH* and SA113 Δ*ptsH*,

respectively, thereby replacing the *ptsH::*lox72 deletion with the *ptsI*::pBT '*ptsI* genomic region containing a functional *ptsHI* operon.

For the construction of *S. aureus ptsH* variants harboring a T to G exchange of nucleotide 136 of the *ptsH* ORF (termed *ptsH*\*), 0.6-kb and 1.1-kb fragments, containing either the promoter region of *ptsH* and the N-terminal part of the *ptsH* ORF (nucleotides 1054446-1054998 of GenBank accession no. AP009351.1) or the C-terminal part of the *ptsH* ORF and an N-terminal fragment of the *ptsI* ORF (nucleotides 1054976-1056121 of GenBank accession no. AP009351.1) were amplified by PCR from chromosomal DNA of *S. aureus* strain Newman using primer pairs MBH-484/MBH-485 and MBH-86/MBH-20, respectively (Supplementary Table S1). Primer MBH-484 contains a non-complementary base that introduces a point mutation in the PCR fragment leading to the T136G exchange of the *ptsH* ORF. Both PCR products were digested with StyI and subsequently ligated with T4 DNA-ligase. The ~1.7-kb ligation product was gel-purified, digested with KpnI/PstI, and cloned into KpnI/PstI-digested pBT to generate plasmid pBT *ptsH1* (Table 1). Presence of the T136G exchange in *ptsH\** harbored by plasmid pBT *ptsH1* was confirmed by sequencing, the plasmid propagated in *E. coli* strain DH5α, electroporated into RN4220 Δ*ptsH*, and selected for tetracycline-resistance. A tetracycline-resistant RN4220 derivative that integrated pBT *ptsH1* at the *ptsHI* locus was used as donor to transduce the *tet*(L)-tagged *ptsH\** allele into the Δ*ptsH* mutants.

*S. aureus* double mutants lacking *ptsH* and *ccpA* were created by transducing the *tet*(L)-tagged *ccpA* deletion of MST14 into Δ*ptsH* derivatives.

#### *2.4. RNA Isolation and Purification, cDNA Synthesis and qRT-PCR*

*S. aureus* strains were cultivated in TSB as described above. Bacterial pellets were collected after 2 h and 8 h of incubation by centrifugation at 5000 rpm at 4 ◦C for 5 min, and immediately suspended in 100 μL ice-cold TE-buffer (10 mM Tris-HCl, 1 mM EDTA, pH 8). Bacteria were disrupted, total RNA isolated, transcribed into cDNA, and qRT-PCRs carried out as described previously [33] using the primers listed in Supplementary Table S1. Transcriptional levels of target genes were normalized against the mRNA concentration of housekeeping gene *gyrB* according to the 2−ΔCT method.

#### *2.5. Measurement of pH, Glucose, Acetate, and Ammonium in Culture Supernatants*

Aliquots (1.5 mL) of bacterial cultures were centrifuged for 2 min at 10,000× *g*, and supernatants were removed, pH measured, and stored at −20 ◦C until further use. Glucose, acetate, and ammonia concentrations were determined with kits purchased from R-Biopharm (Pfungstadt, Germany) and used according to the manufacturer's directions. The metabolite concentrations were measured from at least three independent experiments.

#### *2.6. Biofilm Assays*

Biofilm formation under static conditions was assessed as described [19]. Briefly, overnight cultures were diluted to an OD600 of 0.05 in fresh TSB medium supplemented with glucose to a final concentration of 0.75 % (*w*/*v*), and 200 μL of the cell suspension was used per well to inoculate sterile, flat-bottom 96-well polystyrene microtiter plates (BD). After incubation for 24 h at 37 ◦C without shaking, the plate wells were washed twice with phosphate-buffered saline (pH 7.2) and dried in an inverted position. Adherent cells were safranin-stained (30 sec with 0.1% safranin; Merck, Darmstadt, Germany) and the absorbance of stained biofilms was measured at 490 nm after resolving the stain with 100 μL 30 % (*v*/*v*) acetic acid, using a microtiter plate reader (Victor<sup>2</sup> 1420 Multilabel Counter; Perkin Elmer, Rodgau, Germany).

Biofilm formation under flow conditions was performed as described [34], with minor modifications: Bacteria from overnight cultures were diluted to an OD600 of 0.05 in fresh TSB medium supplemented with glucose to a total concentration of 0.75% (*w*/*v*) and cultivated for 2 h at 37 ◦C with shaking at 150 rpm. Flow cells (Stovall Life Science) were filled with pre-warmed TSB medium supplemented with glucose to a total concentration of 0.75% (*w*/*v*), attached to a peristaltic pump (Ismatec REGLO Digital; Postnova, Landsberg am Lech, Germany) and inoculated with 0.5 mL of the bacterial cultures. Thirty minutes after inoculation, the flowrate was set to 0.5 mL/min and chamber. Biofilm formation was visually documented at different times.

For the assessment of biofilm formation on medical devices under dynamic conditions, peripheral venous catheter (PVC, Venflon Pro Safety 18 G; BD) fragments of 1 cm length were placed into reaction tubes filled with 1 mL of TSB and inoculated with 5 × 105 CFU of TSB-washed bacterial cells obtained from exponential growth phase (inoculation of TSB from overnight cultures to an OD600 of 0.05 and incubation for 2.5 h at 37 ◦C and 225 rpm). The PVC fragments were incubated under non-nutrient limited conditions for five days at 37 ◦C and 150 rpm, and the media were replaced with fresh media every 24 h. PVC fragments were placed five days post inoculation into fresh reaction tubes filled with 1 mL of TSB, biofilms were detached from the catheter surface and resolved by sonification (50 watt for 5 min) followed by 1 min of vortexing. CFU rates and biomasses of resolved biofilms and culture supernatants at day five post inoculation were determined by plate counting and OD600 measurements, respectively.

#### *2.7. Primary Attachment Assay on Polystyrene*

The primary attachment of bacterial cells to polystyrene surfaces was performed as described [35], with minor modifications. Briefly, bacteria from the exponential growth phase (inoculation of TSB from overnight cultures to an OD600 of 0.05 and incubation for 2.5 h at 37 ◦C and 225 rpm) were diluted in TSB to 3000 CFU/mL. 100 μL of the bacterial inoculum was poured onto polystyrene petri dishes (Sarstedt, Nümbrecht, Germany) and incubated under static conditions at 37 ◦C for 30 min. After incubation, petri dishes were rinsed gently three times with 5 mL of sterile PBS (pH 7.5), and subsequently covered with 15 mL of TSB containing 0.8% agar maintained at 48 ◦C. Plates were incubated at 37 ◦C for 24 h. Bacterial attachment to polystyrene was defined as the number of CFU remaining on the petri dish bottom after washing compared to the number of CFU remaining on the petri dish bottom without washing.

#### *2.8. Animal Models*

All animal experiments were performed with approval of the local State Review Board of Saarland, Germany (project identification codes 60/2015 [approved 21.12.2015], and 34/2017 [approved 09.11.2017]), and conducted following the national and European guidelines for the ethical and human treatment of animals. PBS-washed bacterial cells obtained from exponential growth phase cultures were used as inoculum.

For the murine abscess model, infection of animals was carried out as described [33], with minor modifications; specifically, 8- to 12- week-old female C57BL/6N mice (Charles River, Sulzfeld, Germany) were anesthetized by isoflurane inhalation (3.5%; Baxter, Unterschleißheim, Germany) and 100 <sup>μ</sup>L bacterial suspension containing 5 × 107 CFU were administered intravenously by retro bulbar injection. Immediately after infection, animals were treated with a single dose of carprofen (5 mg/kg; Zoetis, Berlin, Germany). Behavior and weight of mice was monitored daily, and four days post-infection, mice were sacrificed, and livers and kidneys were removed. The bacterial loads in liver and kidney tissues were determined by homogenization of weight-adjusted organs in PBS (pH 7.4), followed by serial dilutions on sheep blood agar plates and plate counting after 24 h incubation at 37 ◦C.

For the *S. aureus* based murine foreign body infection model, implantation of catheter fragments and infection of animals was carried out as described [36], with minor modifications: 8- to 12- week-old female C57BL/6N mice (Charles River) were anesthetized by intraperitoneal injection of 0.05 mg/kg body weight fentanyl (Hexal, Holzkirchen, Germany), 5 mg/kg midazolam (Hameln Pharma Plus, Hameln, Germany) and 0.5 mg/kg medetomidine (Orion Pharma, Hamburg, Germany). After treatment with a dose of carprofen (5 mg/kg, Zoetis), the animals were shaved with an animal trimmer (BBraun, Melsungen, Germany) and depilated with asid-med hair removal cream (Asid Bonz, Her-

renberg, Germany) on both flanks. The depilated skin was disinfected with ethanol (70%) and 1 cm catheter fragments (PVC, 14G, Sarstedt) were implanted subcutaneously and inoculated with 1 × <sup>10</sup><sup>4</sup> CFU of the respective *S. aureus* strains. Wounds were closed with staples (Fine Science Tools, Heidelberg, Germany) and anesthesia was antagonized with 1.2 mg/kg body weight naloxone (Inresa, Freiburg im Breisgau, Germany), 0.5 mg/kg flumazenil (Inresa) and 2.5 mg/kg atipamezole (Orion Pharma). Behavior and weight of the animals was monitored daily. Ten days post infection, animals were sacrificed, edema sizes were measured and photo documented, and catheter fragments with surrounding tissue were harvested for microbial analyses. Excised tissues were homogenized in 1 mL TSB with a hand disperser (POLYTRON PT 1200 E; Kinematica, Eschbach, Germany), and biofilms were detached from the PVC fragments and resolved by sonification (50 watt for 5 min) followed by vortexing (1 min). CFU rates in tissue and of biofilm formed on the catheter were determined by plating serial dilutions on sheep blood agar plates and plate counting after 24 h of incubation at 37 ◦C.

#### *2.9. Statistical Analyses*

The statistical significance of changes between groups was assessed by one-way ANOVA followed by Holm-Sidak's post-hoc tests for experiments containing ≥ 5 biological replicates using the GraphPad software package Prism 6.01 (San Diego, CA 92108, USA). *p* values < 0.05 were considered statistically significant.

#### **3. Results and Discussion**

#### *3.1. Growth, pH Characteristics, and Metabolite Profiles Differ between ptsH and ccpA Mutants*

To determine if inactivation of *ptsH* in *S. aureus* leads to changes in growth and carbon catabolism, mutants were constructed in the *S. aureus* laboratory strain Newman (Table 1) and growth and physiology were assessed (Figure 1). In detail, a mutant lacking *ptsH* (Δ*ptsH*) was constructed and cis-complemented (*ptsH::ptsH*), and a Δ*ccpA ptsH* double mutant was created (*ccpA\_ptsH*). In addition, a *ptsH* mutant harboring a point mutation in the *ptsH* gene (T136G) leading to the substitution of serine to alanine at position 46 of HPr (HPr-S46A) was constructed (*ptsH*\*). The phosphorylation at this amino acid represents a known prerequisite for HPr to activate CcpA in other gram-positive bacteria (14), while its activity in the phosphotransferase uptake system (PTS) should be unaffected. The parental strain Newman and the cis-complemented *ptsH* derivative displayed similar growth characteristics and comparable generation times, respectively. In contrast, all *ptsH* mutants (*ptsH*, *ccpA\_ptsH*, and *ptsH*\*) had reduced growth rates in the exponential (1–3 h) and the transition phase (4–6 h) relative to the wild type (Figure 1 and Table 2). Interestingly, the growth rate of the isogenic *ccpA* deletion mutant was only slightly diminished relative to that of strain Newman (Figure 1 and Table 2), and differed significantly from the wild type only during the transition phase. After 12 h of cultivation, growth yields were comparable for all strains (Figure 1b), suggesting that neither the lack of CcpA nor HPr has a clear long-term effect on biomass production of *S. aureus* cultured in rich medium. This is in line with earlier findings regarding CcpA [15].

**Table 2.** Generation time of *S. aureus* strains cultivated in TSB under aerobic conditions.


<sup>1</sup> Data are presented as mean ± SD (*<sup>n</sup>* = 6). <sup>2</sup> *<sup>p</sup>* values were determined by one-way ANOVA and Holm-Sidak's multiple comparison test.

**Figure 1.** Impact of *ptsH* and/or *ccpA* on growth and pH profiles of *S. aureus* TSB cultures. Bacteria were inoculated to an OD600 of 0.05 in TSB and cultured aerobically at 37 ◦C and 225 rpm. OD600 (**a**,**b**) and pH measurements (**c**) of the culture media were determined hourly. Symbols represent: strains Newman (black symbols), Nm *ptsH* (white symbols), Nm *ptsH::ptsH* (grey symbols), Nm *ptsH*\* (yellow symbols), Nm *ccpA* (red symbols), and Nm *ccpA\_ptsH* (blue symbols). The results are the mean ± SD of at least five independent experiments. (**b**) OD600 readings of the cell cultures at 2, 5, and 12 h of growth, respectively. The data are presented as box and whisker plot showing the interquartile range (25–75%, box), the median (horizontal line), and the standard deviation (bars) of 5–6 independent experiments. \*\*, *p* < 0.01 (one-way ANOVA and Holm-Sidak's multiple comparison test. Only differences between Newman and mutants are shown).

To assess the bacterial acid production during growth, the pH of culture supernatants was measured over time (Figure 1c). The pH of culture supernatants from wild type and cis-complemented *ptsH* mutant cultures were similar. In contrast, pH values of culture supernatants from the *ptsH* and *ccpA\_ptsH* deletion mutants indicated that little acid was produced during growth. The pH profiles of the *ptsH*\* and *ccpA* mutant cultures were between these two extremes but indicated that acidic end-products were produced and consumed during growth. Taken together, these data indicate that inactivation of *ptsH*, or interference with HPr phosphorylation, delays growth and medium acidification to a greater extent than does deletion of *ccpA*. In addition, the small differences in physiological parameters (i.e., growth and pH kinetics) between the *ptsH* and the *ccpA\_ptsH* mutants and between the *ccpA* and *ptsH*\* mutants indicate additional, CcpA-independent functions of HPr.

To get an idea about the metabolic processes that are active in Newman wild type and mutant cells cultured in TSB, the concentrations of glucose, acetate, and ammonia were determined in culture supernatants over time (Figure 2). Strain Newman and the cis-complemented *ptsH* mutant (*ptsH::ptsH*) depleted all available glucose in the medium within the first 5 h of cultivation (Figure 2a). In contrast, glucose depletion in Δ*ptsH* and Δ*ccpA\_ptsH* mutant cultures was severely delayed, and low concentrations of glucose were still detectable in culture supernatants even after 10 h of growth. In Δ*ccpA* and *ptsH\**

mutant cultures, glucose levels decreased slower than in wild type cultures, and no glucose was detectable after 7 h of growth (Figure 2a).

**Figure 2.** Impact of *ptsH* and/or *ccpA* on glucose consumption, acetate and ammonia production of *S. aureus* Newman during in vitro growth. *S. aureus* strains Newman (black symbols), Nm *ptsH* (white symbols), Nm *ptsH::ptsH* (grey symbols), Nm *ptsH*\* (yellow symbols), Nm *ccpA* (red symbols), and Nm *ccpA\_ptsH* (blue symbols) were cultivated in TSB, and glucose (**a**), acetate (**b**), and ammonia (**c**) concentrations in culture supernatants were determined hourly. Results are presented as the average and standard deviation of at least three independent experiments.

When *S. aureus* is cultured aerobically in a glucose-containing medium, cells produce and secrete acetate as long as glucose is available [37]. Consistent with this fact, increasing acetate concentrations in the culture supernatants were observed during the first 5–6 h of growth for all strains (Figure 2b). However, while all strains accumulated acetate in the medium, the maximum concentrations differed; specifically, the wild type and the cis-complemented *ptsH* derivative accumulated up to 22 mM of acetate. In contrast, supernatants from Δ*ptsH* and Δ*ccpA\_ptsH* mutant strain cultures had approximately onethird of the concentration of that from the wild type strain. Similar to that seen in the pH profiles (Figure 1c), the acetate profiles of *ptsH*\* and Δ*ccpA* mutant cultures centered in between those two extremes (Figure 2b). At 7–8 h post inoculation, acetate levels decreased in the supernatants of all cultures, irrespective of the fact that glucose was present in Δ*ptsH* and Δ*ccpA\_ptsH* mutant cultures (Figure 2a).

*S. aureus* also utilizes amino acids as carbon sources for growth, a process that requires deamination of the amino acids, resulting in the secretion of ammonia into the culture supernatant [20]. The uptake and catabolism of amino acids in *S. aureus* is subject to CCR [20]. While glucose was present in the medium, ammonia levels remained low in the wild type and *ptsH::ptsH* culture supernatants, followed by a steady increase in the ammonia concentrations (Figure 2c). In contrast, cultures of strains Nm *ptsH* and Nm *ccpA\_ptsH* began to accumulate ammonia beginning at 3 h of cultivation. Interestingly, the ammonia concentration in the supernatant of the Δ*ccpA* mutant closely resembled that of the *ptsH* deletion mutants, while the *ptsH*\* mutant resembled the late induction of the wild type strain (Figure 2c).

Taken together, these data show that the inactivation of *ptsH* or *ccpA* results in distinct differences in glucose consumption, acetate accumulation and reutilization, and ammonia secretion in *S. aureus*. Furthermore, the exchange of an amino acid critical for the interaction of HPr with CcpA in the *ptsH*\* mutant resulted in metabolite profiles (i.e., glucose and acetate) comparable to the Δ*ccpA* mutant, while some alterations in the growth profile, generation time, and ammonia secretion were observed. Importantly, after 12 h of growth, the biomass of *S. aureus* Newman was independent of *ccpA* and *ptsH*, suggesting that *S. aureus* has other means to utilize carbon sources in the growth medium. Specifically, *ptsH* mutants were able to utilize glucose from the growth medium—although much slower than the wild type—demonstrating that *S. aureus* can transport glucose independent of the group translocation PTS [38]. A likely compensatory transporter would be one of the many ATP binding cassette transporters identified in *S. aureus* [39].

#### *3.2. Inactivation of ptsH and/or ccpA Alters Transcription of TCA Cycle and Virulence Factor Genes*

CcpA is known to affect transcription of a large number of central carbon metabolism and virulence genes [15–17]. For this reason, the effect of *ptsH* deletion on transcription of genes regulated by CcpA such as *citB* (encoding the TCA cycle key enzyme aconitase), *pckA* (encoding the gluconeogenesis key enzyme phosphoenolpyruvate carboxykinase), and *hla* (encoding α-hemolysin) was assessed. Specifically, mRNA levels were determined in cells from the exponential (i.e., 2 h) and post-exponential growth phases (i.e., 8 h) by qRT-PCR (Figure 3).

**Figure 3.** Effect of *ptsH* and/or *ccpA* mutations on the transcription of *S. aureus*. Newman wild type and mutant cells were cultured aerobically in TSB, as outlined in Materials and Methods. Cells were harvested at the time points indicated, total RNAs isolated, and qRT-PCRs performed for *citB* (**a**), *pckA* (**b**), and *hla* (**c**). Transcripts were quantified in reference to the transcription of gyrase B. Data are presented as mean + SD of five biological replicates. \*, *p* < 0.05; \*\*, *p* < 0.01 (one-way ANOVA and Holm-Sidak's multiple comparison test; only differences between Newman and mutants are shown).

Consistent with our previous observation that *S. aureus* transcription of *citB* and *pckA* is repressed by CcpA when cultivated with glucose [16], deletion of *ccpA* significantly increased the level of *citB* and *pckA* mRNA in exponential growth phase cells relative to wild type cells (Figure 3a,b). As expected, in the post-exponential growth phase, comparable *citB* and *pckA* transcript levels were observed in the Δ*ccpA* mutant and the wild type. Similar to the Δ*ccpA* mutant, exponential growth phase cells of the *ptsH* mutants (Δ*ptsH*, *ptsH*\*, and Δ*ccpA\_ptsH*) had comparable *citB* and *pckA* mRNA levels, while the cis-complemented *ptsH* derivative (*ptsH::ptsH*) had *citB* and *pckA* transcript levels comparable to the wild type strain. In contrast to exponential growth phase cultures, all three *ptsH* mutants produced significantly lower levels of *pckA* mRNA than the wild type at 8 h, suggesting that *pckA* transcription is affected by HPr at later growth stages in a way that is independent of CcpA. This differed from the results for *citB* in which all *ptsH* mutants (Δ*ptsH*, *ptsH*\*, and Δ*ccpA\_ptsH*) had comparable transcript levels to that of the Δ*ccpA* mutant and the wild type after 8 h of growth. The fact that the *ptsH*\* mutant produced *pckA* transcript levels similar to the *ptsH* mutant, but not similar to the *ccpA* mutant suggests that HPr phosphorylated at serine 46 acts in part independent of CcpA. This observation is consistent with that found in other gram-positive bacteria, where the serine 46-phosphorylated HPr exerted effects on CCR via CcpA and inducer exclusion [40]. However, it cannot be excluded that differences in *pckA* transcription between the *ptsH*\* and Δ*ccpA* mutants are due to differences in protein stability. The reason why protein stability cannot be excluded is because phosphorylation of the *B. subtilis* HPr homolog at Ser-46 stabilized the protein [41], while a serine to alanine exchange of Ser-46 in the *E. coli* HPr homolog was found to decrease the stability of the protein [42].

CcpA represses transcription of *hla* during the exponential growth phase when bacteria are cultured in presence of glucose [15,16]. Similarly, levels of *hla* mRNA from the exponentially growing Δ*ccpA* mutant and all *ptsH* mutants (Figure 3c) were de-repressed,

while the cis-complemented *ptsH* deletion mutant produced *hla* transcript levels that were comparable to the wild type. During the post-exponential growth phase, only the *hla* transcript levels of the Δ*ptsH* and the Δ*ccpA\_ptsH* double mutant were significantly increased (Figure 3c), suggesting that HPr affects expression of α-hemolysin in a CcpA-dependent and -independent manner. Taken together, these data suggest that exponential growth phase *S. aureus* was cultured in the presence of glucose, HPr affects the transcription primarily via activation of CcpA, while in the post-exponential growth phase cells of *S. aureus*, HPr is likely to affect gene transcription by CcpA-independent mechanism(s).

#### *3.3. Impact of ptsH Deletion on Biofilm Formation of S. aureus SA113*

CcpA is important for polysaccharide intercellular adhesin (PIA)-dependent biofilm formation by staphylococci under glucose-rich in vitro conditions [19,24,43]. The importance of HPr on sugar import and gene regulation suggests that HPr might influence biofilm formation of *S. aureus*. Strain Newman is a weak biofilm producer in glucose-rich medium under in vitro conditions [34], hence we transduced the *ptsH* mutations into *S. aureus* strain SA113, which forms a strong biofilm under these conditions [19,22]. The ability of SA113 mutant strains were analyzed using a semi-quantitative static biofilm assay (Figure 4a) and in biofilm flow cells (Figure 4b).

**Figure 4.** Mutations of *ptsH* affect biofilm formation of *S. aureus* under static and flow conditions. (**a**) Biofilm growth of *S. aureus* strains in a static 96-well microplate assay. The data show the mean + SD of five biological replicates. (**b**) Flow cell chambers were inoculated with *S. aureus* strains as indicated, allowed to attach to the surfaces for 30 min, and incubated under constant flow for 24 h. The results shown are representative of two independent experiments. (**c**) Effect of the *ptsH* mutation on the transcription of *icaA* in *S. aureus* strain SA113. Cells of SA113, the Δ*ptsH* mutant, and the cis-complemented *ptsH::ptsH* derivative were cultured aerobically in TSB. After 2 h of growth, cells were harvested, total RNAs isolated, and qRT-PCRs performed for *icaA*. Transcripts were quantified in reference to gyrase B mRNA. Data are presented as mean + SD of five biological replicates. \*\*, *p* < 0.01 (one-way ANOVA and Holm-Sidak's multiple comparison test; only differences between SA113 and mutants are shown).

Under static conditions, the Δ*ptsH* and Δ*ccpA\_ptsH* mutants of SA113 displayed drastic decreases in their biofilm formation capacities on polystyrene surfaces, whereas the cis-complemented derivative (*ptsH::ptsH*) formed biofilms that were comparable to the ones seen with the wild type (Figure 4a). Deletion of *ccpA* or the S46A mutation of HPr in SA113 (*ptsH*\*) also significantly reduced biofilm formation, however, not to the extent seen with the Δ*ptsH* mutant, supporting our hypothesis that a functional PTS is important for *S. aureus* to form a biofilm in this type of assay. In the flow chamber assay, the Δ*ptsH* mutant failed to produce a clear biofilm within the microchannel after 24 h of constant flow, while both, the wild type and the cis-complemented *ptsH* derivative, almost completely filled the microchannel with biomass (Figure 4b), suggesting that HPr is also important for biofilm formation under shear flow. To exclude that the latter phenotype

was caused by a decreased capacity of the Δ*ptsH* mutant to attach to the microchannel surface, the primary attachment capacities of the strains were determined. Here, no clear differences in attachment towards polystyrene surfaces were obtained for the strain triplet, suggesting that the observed lack of biofilm formation of the SA113 Δ*ptsH* mutant is likely due to a deficiency in biofilm maturation. To determine whether this effect might be due to a decreased capacity of the mutant to produce PIA, we assayed the transcription of *icaA*, which is part of the *icaADBC* polycistronic mRNA that encodes proteins needed for PIA synthesis [22]. Consistent with the reduced ability of the SA113 *ptsH* mutant to form a biofilm under static and flow conditions, we observed significantly decreased levels of *icaA* transcripts in the *ptsH* deletion mutant relative to the wild type and the ciscomplemented mutant (Figure 4c). Together, these data suggest that HPr, in part, promotes biofilm formation of *S. aureus* by enhancing the expression of the PIA synthesis machinery.

In a third biofilm assay intended to resemble the in vivo situation more closely, we studied the ability of SA113 and its derivatives to form biofilms on peripheral venous catheter (PVC) fragments under non-nutrient limited conditions (Figure 5).

**Figure 5.** Inactivation of *ptsH* and/or *ccpA* reduces the biofilm formation capacity of *S. aureus* on medical devices. (**a**) Images of *S. aureus*-loaded catheter fragments at day 5 post inoculation (6.3-fold magnification). The results are representative of three independent experiments. (**b**,**c**) Colony forming units (CFU) and total biomass of detached biofilms were determined by plate counting (**b**) and measuring the OD600 of the TSB solutions (**c**). The data are presented as box and whisker plot showing the interquartile range (25–75%, box), the median (horizontal line), and the standard deviation (bars) of nine independent experiments. \*\*, *p* < 0.01 (one-way ANOVA and Holm-Sidak's multiple comparison test; only differences between SA113 and mutants are shown).

Using this assay, a strong biofilm was macroscopically detectable on catheter fragments inoculated with the wild type or the cis-complemented *ptsH* derivative at 5 days post inoculation (Figure 5a). In contrast, on catheter fragments inoculated with either the Δ*ptsH* mutant, the Δ*ccpA* mutant, the *ptsH*\* mutant, or the Δ*ccpA\_ptsH* double mutant, almost no biofilm was visible. These observations were further supported by CFU and OD600 determinations of TSB solutions harboring the detached biofilms (Figure 5b,c). A significant reduction in viable bacteria (~1 log) was observed on all fragments inoculated with mutants when compared to the wild type inoculated fragments (Figure 5b). Similar to the CFU data, the OD600 values were approximately 10-fold lower in the detached biofilms formed by the mutants (Figure 5c), suggesting that *ptsH* and *ccpA* deletions elicit rather comparable effects on the biofilm forming capacity of *S. aureus* on PVC surfaces under non-nutrient limited conditions.

#### *3.4. HPr Contributes to Infectivity and Biofilm Formation of S. aureus SA113 in a Murine Foreign Body Infection Model*

CcpA is not required for biofilm formation of *S. aureus* and *S. epidermidis* on implanted catheter fragments in normoglycemic mice [24,43], but this did not address the function of HPr in vivo. In order to address this question, the ability of strains SA113, the Δ*ptsH* mutant, and the cis-complemented *ptsH* derivative to form biofilms on implanted catheter fragments was assessed in the murine foreign body infection model [36] with normoglycemic mice (Figure 6).

**Figure 6.** Inactivation of *ptsH* reduces the infectivity of *S. aureus* SA113 in a murine foreign body infection model. Catheter fragments were implanted subcutaneously into the back of normoglycemic mice and inoculated with cells of *S. aureus* strains SA113 (black symbols), its *ptsH* deletion mutant (white symbols), and the cis-complemented *ptsH* mutant (grey symbols), respectively (*n* = eight animals per group). Ten days post infection, animals were euthanized, edema sizes around the implanted catheters were measured (**a**), and the catheters and surrounding tissues were explanted. Bacterial loads from catheter detached biofilms (**b**) and in surrounding tissue homogenates (**c**) were determined by CFU counting. The data represent the values of every individual animal (symbols) and the median (horizontal line). \*\*, *p* < 0.01 (one-way ANOVA and Holm-Sidak's multiple comparison test).

Mice challenged with the Δ*ptsH* mutant displayed a clear reduction in edema sizes around the implanted catheter fragments (Figure 6a) and a small but significant reduction (~2-fold) in detached bacteria (Figure 6b), when compared to animals infected with wild type bacteria or the cis-complemented *ptsH* derivative. In contrast, no significant differences in bacterial loads of tissues surrounding the catheter fragments were obtained (Figure 6c). These findings suggest that HPr, unlike CcpA, has a small but important function on the biofilm formation capacity of PIA producing *S. aureus* in normoglycemic mice. This difference is probably due to a reduced sugar uptake capacity of the *ptsH* mutant, which might interfere with the enhanced carbon and energy demand of *S. aureus* during biofilm maturation.

#### *3.5. HPr and CcpA Are Both Required for Full Infectivity of S. aureus in a Murine Liver Abscess Model*

The formation of liver abscesses is one of the clinical manifestations caused by *S. aureus* in which CcpA exerts a strong effect on disease progression in normoglycemic mice [6]. To determine how P-Ser-HPr affects infectivity of *S. aureus* in a murine liver abscess model, the bacterial loads in livers four days post infection were assessed (Figure 7).

Consistent with previous observations [6,24], we observed a nearly 3 log reduction in bacterial loads in liver tissue of C57BL/6 mice challenged with the *ccpA* mutant bacteria (median 2.2 × <sup>10</sup><sup>5</sup> CFU/g tissue) relative to mice infected with the wild type strain (median 7.6 × <sup>10</sup><sup>7</sup> CFU/g tissue). Importantly, a greater reduction in CFU/g liver was observed (~4 log; median 1.4 × 104 CFU/g tissue), when mice were challenged with the <sup>Δ</sup>*ptsH* mutant. Infection of mice with the cis-complemented *ptsH* derivative resulted in a bacterial burden in the liver (median 4.2 × <sup>10</sup><sup>7</sup> CFU/g tissue) comparable to that seen in wild type infected mice, demonstrating that the decreased CFU rates determined in liver tissues of Δ*ptsH* infected mice were due to the lack of HPr. Notably, mice challenged with the *ptsH*\* mutant carrying the S46A exchange in HPr also caused an almost ~4 log reduction in bacterial loads in liver tissue (median 2.1 × 104 CFU/g tissue), suggesting that both, the deletion of *ptsH* and a mutation of serine 46 of HPr, alter the virulence of *S. aureus* in this murine infection model in a CcpA-independent manner. The lowest CFU rates in liver tissues were observed when mice were challenged with the Δ*ccpA\_ptsH* double mutant (median 4.8 × 103 CFU/g tissue), suggesting that CcpA might also exert some effects on virulence of *S. aureus* in this infection model independently of HPr.

**Figure 7.** Inactivation of *ptsH* and/or *ccpA* results in decreased bacterial burden of *S. aureus* Newman in a murine abscess model. C57BL/6N mice were infected retro bulbar with 5 <sup>×</sup> <sup>10</sup><sup>7</sup> CFU of *S. aureus* strains Newman (black symbols), its Δ*ptsH* (white symbols) and Δ*ccpA* (red symbols) mutants, the *ptsH*\* mutant (yellow symbols), the Δc*cpA\_ptsH* double mutant (blue symbols), and the cis-complemented *ptsH* derivative (grey symbols), respectively. Bacterial loads in liver tissue homogenates were determined four days post infection. The data display the median (horizontal lines) and individual values of every animal (dots; *n* = 8–10 animals per group). \*\*, *p* < 0.01 (one-way ANOVA and Holm-Sidak's multiple comparison test; only differences between Newman and mutants are shown).

#### **4. Conclusions**

Central carbon metabolism and virulence factor synthesis are tightly linked in *S. aureus* and controlled by several transcription factors [3]. Notably, CcpA is the only transcription factor known to enhance infectivity of *S. aureus* [6,24,25], while other regulators such as CcpE, CodY, and RpiRc are thought to attenuate rather than to promote infectivity of this bacterium in mice [33,44–47]. We show here that HPr contributes positively to infectivity of *S. aureus* in mice, presumably by affecting central carbon metabolism and virulence factor synthesis in a CcpA-dependent and -independent manner. These effects are likely mediated through changes in sugar transport and carbon metabolization that alter biofilm formation [24]. It is also possible that HPr in *S. aureus* acts like the HPr homolog of *E. coli* to modulate quorum sensing by interacting with autoinducer-2 (AI-2) modifying factors [48]. Given the importance of HPr on biofilm formation and virulence in *S. aureus*, this phosphocarrier protein could be a promising drug target for the development of novel anti-staphylococcal compounds.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/2076-260 7/9/3/466/s1, Table S1: Primers used in this study.

**Author Contributions:** Conceptualization, M.B., R.G. and G.A.S.; methodology, L.P., E.-L.B. and R.G.; investigation, L.P., A.-C.B., E.-L.B. and L.Z.; writing—original draft preparation, L.P. and M.B.; writing—review and editing, M.B., R.G. and G.A.S.; visualization, M.B. and R.G.; supervision, R.G.; project administration, M.B. and R.G.; funding acquisition, M.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the German Research Foundation (DFG), grant numbers BI 1350/1-2 and SFB1027. We acknowledge the support by the DFG and Saarland University within the funding program Open Access Publishing.

**Data Availability Statement:** The datasets generated and analyzed during the current study are available from the corresponding author on reasonable request.

**Acknowledgments:** The authors are grateful to Karin Hilgert and Benjamin Kastell for excellent technical assistance.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

