Novel lipophosphonoxin-loaded polycaprolactone electrospun nanofiber dressing reduces Staphylococcus aureus induced wound infection in mice

Active wound dressings are attracting extensive attention in soft tissue repair and regeneration, including bacteria-infected skin wound healing. As the wide use of antibiotics leads to drug resistance we present here a new concept of wound dressings based on the polycaprolactone nanofiber scaffold (NANO) releasing second generation lipophosphonoxin (LPPO) as antibacterial agent. Firstly, we demonstrated in vitro that LPPO released from NANO exerted antibacterial activity while not impairing proliferation/differentiation of fibroblasts and keratinocytes. Secondly, using a mouse model we showed that NANO loaded with LPPO significantly reduced the Staphylococcus aureus counts in infected wounds as evaluated 7 days post-surgery. Furthermore, the rate of degradation and subsequent LPPO release in infected wounds was also facilitated by lytic enzymes secreted by inoculated bacteria. Finally, LPPO displayed negligible to no systemic absorption. In conclusion, the composite antibacterial NANO-LPPO-based dressing reduces the bacterial load and promotes skin repair, with the potential to treat wounds in clinical settings.


Results
Wound dressing preparation and morphology. PCL-based nanofibrous wound dressings were prepared by blending electrospinning technology 17,18 , with the addition of increasing concentrations (0, 2, 5, and 10 wt%) of LPPO DR-6180 (hereafter LPPO). LPPO was selected for its strong antibacterial properties and low cytotoxicity and was synthesized according to the procedure published previously 6 . The quality of the prepared nanofibrous materials was assessed by scanning electron microscopy (SEM). SEM of scaffolds with subsequent basic image analysis of fiber diameters revealed comparable morphology of prepared materials (Fig. 1). Histograms show that the fiber diameters were 579 ± 60 nm for PCL, 469 ± 44 nm for NANO-LPPO2%, 664 ± 56 nm for NANO-LPPO5%, and 608 ± 70 nm for PCL-LPPO10%.
LPPO release in PBS and LPPO release during enzymatically catalyzed degradation. For the antibiotic function of the NANO-LPPO wound covers to be effective, controlled release of the active compound, LPPO, is necessary. Figure 2A shows a weight loss of NANO-LPPO samples incubated in PBS in the presence or absence of lipase. In a phosphate-buffered saline environment, only the material with the highest concentration of LPPO (NANO-LPPO10%) showed a strong release of LPPO (Fig. 2B). The LPPO release from NANO-LPPO10% occurred gradually, in the first 24 h 2.9 mg was released rapidly. This is the so-called burst release, which is typical for drugs incorporated using blending electrospinning technology 17,18 . Then the release continued more slowly and after 7 days the total amount of LPPO released was 3.3 mg. For NANO-LPPO2%, no detectable drug release was observed during the 7-day incubation in PBS.
We next monitored the LPPO release in the presence of a lipase. Lipases are ubiquitous enzymes capable of degradation of PCL-based nanomaterials present in humans and/or produced by bacteria. The lipase-catalyzed degradation assay (experiments B1-B3, Table 1) indicated that LPPO was released during NANO degradation in a concentration-dependent manner (Fig. 2C). Importantly, the release occurred even from NANO-LPPO2%, which was resistant to release in PBS. The release followed a two-step kinetic, with the most pronounced release during the first day (75%), followed by a reduced degradation rate during the remaining observation period. The total amount of LPPO released from 50 mg of the original sample (NANO-LPPO2%) was 0.75 mg. Interestingly, the NANO-LPPO10% was completely degraded within 24 h (50% was lost within the first 8 h). The total amount of LPPO released during the experiment from 50 mg of the original total sample weight of NANO-LPPO10% was 3.39 mg.
The level of degradation was also reflected by distinct changes in the fiber morphology ( Fig. 2D-F). When compared to the control sample (NANO without LPPO) (Fig. 2D) where structural changes occurred only on the surface of fibers, NANO-LPPO2% and NANO-LPPO10% (Fig. 2E,F) displayed more advanced stages of material degradation with broken fibers probably due to complete splitting of the polymer in its amorphous parts. The NANO-LPPO2% scaffolds were also characterized by the formation of web-like fiber structures (Fig. 2E), not detected in the NANO-LPPO10% material. www.nature.com/scientificreports/ the wettability of the nanofiber (Fig. S1). The Washburn method 19,20 revealed faster initial weight gain (higher wicking rate) for PCL loaded with LPPO (NANO-LPPO10%). The final slope of line at y = 0.215 ± 0.003 g 2 s −1 and regression coefficient at R = 0.83 ± 0.07 were measured for NANO-LPPO10% samples. On the other hand, the slope of the line for the negative control was significantly smaller at A = 0.0003 ± 0.00015 g 2 s −1 with the regression coefficients at R = 0.96 ± 0.03.

Effect of LPPO on HDFs and HaCaTs.
Since the antibacterial activity of the NANO-LPPO wound dressing is based on the controlled release of LPPO from the nanofibrous scaffold, we were interested in whether LPPO at bactericidal concentrations modulates proliferation/differentiation of fibroblasts and keratinocytes. Furthermore, we also investigated the ability of LPPO to interfere with TGF-1 signaling that is important for proper wound closure 21 .
Effect of LPPO on cell proliferation. The in vitro investigation conducted on HDFs (Fig. 3A) and HaCaT ( Fig. 3B) cells revealed that LPPO in the concentration range from 0.1 to 25 mg/L (bactericidal concentrations) does not impair the metabolic activity of studied cells. Of note, higher tested concentrations of LPPO, 50 and 100 mg/L, were found toxic to studied cells.
Effect of LPPO on cell migration. It is well known that keratinocyte migration is essential for proper wound reepithelization, thus we examined how LPPO affects 2D wound healing in HaCaT cells. As illustrated in Fig. 4 keratinocytes migration was accelerated in the presence of TGF-1β (positive control). When compared to the control, LPPO (at any of the tested concentrations) did not significantly change the migratory ability of cells. However, LPPO attenuated TGF-1β-induced migration at higher tested concentrations of 5 and 10 mg/L, but not when added at 1 mg/L.   Fig. 5. As observed in HDFs (Fig. 5A), TGF-β1 activated the canonical (SMAD) signaling resulting in stimulated production of SMA and fibronectin. Similarly, TGF-β1 increased pAKT expression whereas the expressions of pERK/ERK were rather slightly decreased. HDFs treatment with LPPO, in the presence or absence of TGF-β1, slightly decreased fibronectin and SMA expressions in a concentration-dependent manner. In this context, the TGF-β1-induced canonical signaling was also slightly attenuated with increasing concentration of LPPO. A similar picture was seen in the trend of pAKT expression while total AKT was rather stimulated with LPPO. On the other hand, LPPO activated the non-canonical (pERK) signaling, but this effect was abolished in the presence of TGF-β1. In parallel, HaCaT phenotype ( Fig. 5B) was characterized by no production of fibronectin, expression of AKT/ ERK total forms (absence of pAKT, poor expression of pERK), marked expression of keratins-8, -14, and -19 as well as the presence of E-cadherin and absence of N-cadherin. TGF-β1 treatment of keratinocytes resulted in fibronectin expression unchanged pAKT and decreased total AKT production as well as increased pERK levels and to E-to N-cadherin switch. The keratin profile remained rather stable. LPPO slightly stimulated pAKT, but reduced total AKT and pERK/ERK levels. Moreover, E-cadherin expression was stimulated at lower tested  Table 1). Immunofluorescence of HDFs and HaCaTs following treatment with LPPO. Immunofluorescence staining of fibronectin showed that HDFs (Fig. 6) form ECM scaffold, which is more fibronectin-dense following TGF-β1 treatment. As expected, TGF-β1 also stimulated differentiation of HDFs into myofibroblasts (with well-formed SMA stress fibers). LPPO did not remarkably reduce fibronectin deposition nor induced SMA expression. Importantly, the TGF-β1-induced ECM and SMA production were not affected by LPPO at 1 and 5 mg/L. Of note, LPPO at 10 mg/L slightly decreased ECM synthesis and TGF-β1-induced ECM and SMA depositions. The morphology of HaCaT cells (Fig. 7) was clearly different in cells treated with TGF-β1 (large cells) when compared to the control (small cells). Untreated cells expressed moderate levels of keratins 14 and 19 while the expression of keratin 8 was minimal. Both, TGF-β1 and LPPO slightly increased keratin 8 levels.
Mice model. Next, we demonstrated that used animal model is suitable to study the clinically relevant S. aureus as wound healing impairing pathogen and NANO-LPPO as an efficient antibacterial wound dressing. The wound experiment was evaluated on post-operative day 7. In detail, wound infection was clearly visible by the presence of purulent exudate (not shown). NANO-LPPO5%-and NANO-LPPO10%-based wound dressings displayed strong infection-impairing properties whereas NANO-LPPO2% exhibited only limited activity. Subsequently, the animal experiment was evaluated microscopically (see "Histology of open wounds" section) as well as using agar-based culture (see "Cultivation and characterization of bacteria" section) and PCR-based   Histology of open wounds. For the histological analysis we used hematoxylin and eosin, Sirius red (collagen) and Gram (gram-positive bacteria) staining; the results are shown in Fig. 8. First, the effect of NANO-LPPO on the healing of non-infected mice was examined. On day seven, the demarcation line (dominantly formed by neutrophils) separated vital tissue from necrosis, thus the acute inflammatory phase was almost finished. The epithelial sheet migrated beneath the scab, but did not reach the central portion of the wounds. The newly formed granulation tissue (GT) was populated by fibroblasts and luminized vessels. However, GT was well-formed only at wound edges whereas the central parts of wounds were filled only with provisional wound matrix. The control NANO dressing slightly improved granulation tissue formation that was seen also in the central parts of wounds while other parameters remained rather stable. The use of NANO-LPPO-based (at any of the tested LPPO concentration) wound dressing did not alter the morphological dynamics of examined parameters.  www.nature.com/scientificreports/ Next, histological analysis of S. aureus infected wounds revealed the known capacity of this bacterium to impair the healing process. In a qualitative comparison to non-infected control, infection led to persisting infiltration of injury site with inflammatory cells, re-epithelization restricted to a thickening of cut edges of the epidermis, remarkably decreased numbers of fibroblasts and luminized vessels with an absence of newly formed collagen in the granulation tissue. A similar picture was detected in wounds treated with LPPO at 2%. Importantly, an increase in re-epithelization, granulation tissue formation, collagen score was observed at this time point for LPPO 5% and 10% with no remarkable differences from the uninfected control.
Cultivation and characterization of bacteria. The next approach was the inspection of bacteria from wound swabs grown on agar media. Agar cultures were first inspected visually, then streaked on agar plates, and the resulting colonies, representing different morphotypes, were identified by MALDI-TOF MS (Table S1). Staphylococcus aureus CNCTC 5480 was detected in all samples without LPPO (S31-S36) or with 2% LPPO (S31-S36). In contrast, samples containing 5% (S43-S48) and 10% (S49-S54) LPPO yielded no visible growth of S. aureus CNCTC 5480, with the only exception of S48. Moreover, visual inspection revealed that samples with 2% LPPO produced reduced growth of S. aureus as compared with the samples without LPPO. Some of the BA plates with either control samples (S1-S30) or inoculated samples containing 5-10% LPPO (S43-S54) yielded sporadic colonies of non-S. aureus bacteria such as Staphylococcus sciuri, Enterococcus gallinarum or unspecified alfa-hemolytic streptococci. All these bacteria are known as potential colonizers of mammalian skin, with S. sciuri being commonly isolated from laboratory mice. In addition, control samples S7-S10 and inoculated samples S37 and S48 yielded colonies of a strain of S. aureus, which was both phenotypically and genotypically distinct from S. aureus CNCTC 5480. The most likely source of the bacteria other than S. aureus CNCTC 5480 was prior skin colonization of experimental mice and/or cross-contamination occurring during the experiment. In any case, the presence of the non-S. aureus CNCTC 5480 bacteria did not interfere with the interpretation of the results. www.nature.com/scientificreports/ Bacteria qPCR-based quantification in wounds. The third approach was qPCR-based quantitation of S. aureus in the wound and surrounding skin. The results are shown in Fig. 9. From the qPCR data, it was apparent that the bacterial load in the wounds covered with NANO-LPPO5% and 10% compared to NANO and NANO-LPPO2% was reduced by several orders of magnitude, correlating with the findings from the histological and bacteriological analyses. We note that qPCR can detect also DNA releases from dead bacteria. Hence, the actual load of viable bacterial cells was likely lower than the apparent values obtained by qPCR.

Determination of LPPO in mouse plasma and liver by LC-MS/MS. Finally, to evaluate whether
LPPO can be after its topical application absorbed and systemically distributed, we analyzed samples of blood and liver homogenates for its presence. As shown in Tables S2 and S3, the amounts of LPPO detected in blood and liver correlated with its amounts in the NANO-LPPO composite. However, the detected systemic LPPO levels were negligible from the cytotoxicity point of view 6 .

Discussion
In the present study, we developed and successfully tested a new electrospun PCL-based scaffold loaded with antibacterial LPPO molecule (NANO-LPPO) promoting healing in a standardized murine model of S. aureus induced wound infection. The biocompatible and biodegradable scaffold consists of a skin ECM-like 22 macroporous network of cross-linked smooth fibers (diameter around 1 µm, pores of several micrometers), with the ability to extract wound exudate and deliver oxygen 23 . This biodegradable material provides inherent adhesion and occlusion, integrates into the wound and reduces the rate of dressing replacement 24 . Scaffolds loading with LPPO did not affect fiber size but enhanced wettability and biodegradation and provided significant antibacterial properties.
The degradation of PCL started as light surface modifications and continued from day 2 in significant restructuring of the entire fiber. In detail, fiber morphology changed due to the rearrangement of macromolecules as a result of the formation of regions with crystalline and amorphous structure 25,26 . Interestingly, restructuring was more pronounced when PCL was loaded with LPPO 27 as demonstrated by crushed/broken fibers due to the complete splitting of the polymer in the amorphous regions. In this context, we may hypothesize at least two basic mechanisms by which LPPO enhances NANO degradation. Firstly, the mechanical properties of NANO could be directly altered by the presence of LPPO molecule. Secondly, LPPO, as an amphiphilic molecule, acts as a surfactant and thus decreases the hydrophobicity of NANO by facilitating the access of the hydrolytic enzymes to a larger surface of the NANO-fiber. Molecular interactions between the hydrophobic parts of NANO and LPPO may lead to changes in the fiber structure enhancing the diffusion of enzymes into the material. Therefore, decreasing hydrophobicity of NANO in the presence of LPPO has been verified via measurement of the wettability. As expected NANO-LPPO10% exhibited significantly better wettability when compared to the control (hydrophobic pure PCL material). www.nature.com/scientificreports/ LPPO release from NANO demonstrated a dose-dependent pattern in PBS. Whereas the release of LPPO10% occurred gradually by the so-called burst release (typical for drugs incorporated using blending electrospinning technology 17,18 ) followed by a decrease of the release rate, LPPO2% was not released during the 7-day incubation period. It appears that the release of LPPO by simple diffusion occurs only at higher concentrations of the incorporated LPPO. This correlates with our previous findings where the elution kinetics of LPPO from polymethyl methacrylate cement (used in orthopedic surgery) resembled the elution of well-known antibiotics 7,28 , supporting the clinical potential of the proposed dressing.
Interestingly, enzymatically catalyzed degradation of NANO revealed a positive correlation of the degradation rate with the amount of loaded LPPO molecule. An initial spike was observed within a few hours after the application, followed by a more gradual prolonged elution in the order of days. This aligns well with the potential therapeutic use, where the elevated initial concentration kills existing bacteria, and this is followed by the elution of decreased, but still sufficient, amounts of the compounds providing the antibacterial milieu for an extended time 29,30 . Moreover, our experiment also indicates that LPPO release is also stimulated by the presence of bacteria in the wound by increasing the lipase activity in the presence of S. aureus 31 .
Current in vitro experiment was conducted to reveal whether LPPO that is released from NANO during the degradation of PCL impairs selected biological processes associated with wound repair. LPPO impaired neither the proliferation of human dermal fibroblasts nor TGF-β1-induced fibroblast-to-myofibroblast transition or fibronectin-rich ECM formation. In particular, the pro-fibrotic cytokine (TGF-β1) is transiently up-regulated in normal skin wounds where its ability to induce myofibroblasts is relevant for wound closure 32 . The LPPO ability to interfere with TGF-β1 signaling, both canonical TGF-β1/SMAD and non-canonical/non-SMAD (MAPK, PI3K) 33 , was not observed in our experiment. We further tested whether LPPO modulates the proliferation and differentiation of keratinocytes as the principal component of the epidermis. The panel of studied keratins (K8, K14, and K19) and significant prevalence of E over N Cadherin revealed the normal status of keratinocytes in the absence/presence of LPPO. The ability of TGF-β1 either to induce epithelial-to-mesenchymal transition 34 (E to N Cadherin switch) or to reduce proliferation 35 was also not modulated by LPPO.
In the in vivo study we used a modified experimental model originally adapted to study antimicrobials for their ability to reduce Acinetobacter baumannii-induced wound infection in mice 36 . We demonstrated that this model is also suitable to study the clinically relevant S. aureus as wound healing impairing pathogen and NANO-LPPO as an efficient antibacterial wound dressing. LPPO exerted topical antibacterial activity when loaded at 5 and 10% with negligible systemic levels in liver and/or plasma. Similarly, a bioactive antibacterial (with sustained amoxicillin release for up to 1 week) bilayer PCL/gelatin nanofibrous scaffold hindered bacterial growth and promoted full-thickness wound healing 37 . Moreover, our newly developed dressing also slightly improved granulation tissue formation and this effect persisted when NANO was loaded with LPPO. The healing promoting effect can even be improved by specific modifications of the material structure, for example by combining the PCL with fish collagen and covalently bonded chitooligosaccharides 38 . Another scaffold comprised two supportive PCL-chitosan layers (accelerating healing) on the sides and a polyvinyl alcohol-metformin hydrochloride (down-regulating expression of fibrosis-related genes) in the middle 39 . From this point of view fabricated scaffolds are promising candidates for treatment (reducing fibrosis/infection and facilitating repair/regeneration) of full-thickness wounds in vivo.

Conclusion
In the present study, we have demonstrated that the PCL-based/LPPO-loaded active wound dressing (NANO-LPPO) is effective in preventing S. aureus-induced wound infection in mice. The in vitro cytotoxicity and morphology assays indicated that the PCL-LPPO nanofibrous material has promising biocompatibility. LPPO is likely released by two main mechanisms, i.e. diffusion from the PCL and degradation of the PCL. LPPO exerted topical antibacterial activity when loaded at 5 and 10%. The most interesting and novel feature of the antibacterial dressing represents the ability of bacteria to increase the LPPO release rate into the wound. In detail, the enzymatic degradation of the nanofibrous scaffold is promoted by LPPO itself and also by (microbial) lipases. In other words the more bacterial burden in the wound will be present the more antibacterial compound will be released. A mechanism that may be found clinically relevant also for other medical applications (e.g. suture material). In summary, PCL-LPPO represents a promising strategy to replace traditional wound dressings with a potential to promote wound repair also in clinical settings. Nevertheless, the use of a murine model represents a limitation of our study due to interspecies variability. Further pre-clinical studies conducted on large animal models physiologically/evolutionary closer to humans are thus warranted by the presented set of data.

Materials and methods
LPPO. The second generation LPPO DR-6180 (Fig. S4) was synthesized in multigram quantities according to the procedure reported previously 6 .
Nanofibrous materials preparation and loading with LPPO. Nanofibrous wound dressings were prepared by blending electrospinning technology 18,40 . Four nanofibrous (polycaprolactone-PCL) materials with different concentrations of LPPO were prepared (0, 2, 5, and 10 wt%). All concentrations were calculated for dry matter (excluded solvent). Firstly, LPPO was dissolved in chloroform/ethanol (Penta s.r.o., Praha, Czech Republic) solvent mixture at a weight ratio of 8:2 (solution for preparation of NANO supplemented with LPPO 2% was prepared by dissolving 0.32 g of LPPO in 84 g of solvent mixture; solution for preparation of PCL supplemented with 5% LPPO was prepared by dissolving 0.8 g of LPPO in 84 g of solvent mixture; solution for preparation of PCL supplemented with 10% LPPO was prepared by dissolving 1.6 g of LPPO in 84 g of solvent mixture). The ultrasound (40 kHz, 50 W) was then applied for 30 min to enhance the solubilization of LPPOs www.nature.com/scientificreports/ and ensure homogeneity of the spinning solutions using an ultrasonic bath (Shesto, Watford, UK). Then 16 g of polycaprolactone (Merck, Praha, Czech Republic) with an average molecular weight M n 45,000 was dissolved in each solution. The polymer concentration used for obtaining continuous fibrous meshes was 16 wt% (LPPO is not included in this ratio). All materials were electrospun using Nanospider NS 1WS500U (Elmarco a.s., Liberec, Czech Republic) on polypropylene spunbond microfiber nonwoven (Pegatex S; 20 g/m 2 ; fiber diameter 20 µm; PFNonwovens Czech, Czech Republic). For electrospinning, a positive voltage was applied to the wire (i.e., the spinning electrode) and a negative voltage was applied to the collector; the potential difference was 40 kV. The distance between the spinning electrode and the collector was 160 mm. The temperature during the experiments was 22 ± 5 °C, with a relative humidity of 40 ± 5%. Produced samples were finally sterilized in the flow box using a UV light for 10 min for each side.
Scanning electron microscopy (SEM) and morphological analysis of PCL. Morphological analysis was performed by scanning electron microscopy (Vega Tescan 3, Tescan, Brno, Czech Republic) after 7 nm of gold sputter coating (Quorum Q50ES, Quorum Technologies, Laughton, Great Britain). Samples were analyzed at an accelerating voltage of 10 kV. The average fiber diameter and the diameter distribution were evaluated using NIS Elements (Laboratory imaging s.r.o., Prague, Czech Republic) using 300 randomly selected and measured fields per material. Histograms were used to summarize data from fiber diameter/distribution measurement.

Wettability of PCL.
The wettability of PCL electrospun nanofibrous materials (supplemented with or without LPPO) was tested by microtensiometer Krüss K121 (Krüss GmbH, Germany). Dynamic wetting was tested by the Washburn adsorption test. Similar surface densities of these samples were used: (1) PCL (NANO) and density of 27.1 ± 1.3 gm 2 and (2) PCL supplemented with 10 wt% of LPPO (NANO-LPPO10%) and density of 28.3 ± 1.8 gm 2 . The sample of 30 mm width and 40 mm length was cut and placed into the holder for foils. The holder with nanofibrous sample was then inserted into the clip, with the bottom sample edge arranged in a horizontal position and parallel orientation of the sample edge during immersion in the test liquid (HPLC water, VWR Int., Stříbrná Skalice, Czech Republic). The ambient conditions were temperature of 22 °C and 58% relative humidity. The wicking, weight of water wicked into the fibrous system in time, into NANO and NANO-LPPO10% was measured five times for each condition. The spontaneous wicking of porous materials even of textiles or nanofibrous layers can be described by the Washburn equation 20 . The results were expressed in the form of average squared mass gain over time at the beginning of wicking represented by the direction of the regression curve of the wicking rate (Fig. S1).

In vitro study in human dermal fibroblasts (HDF) and human keratinocyte cell line (HaCaT).
HDFs were isolated from skin specimens, as previously described 41 , obtained from two healthy donors during aesthetic surgery with the informed consent of patients (in complete agreement with the Helsinki Declaration after approval by the Ethical Committee of the University Hospital Královské Vinohrady) at the Department of Aesthetic Surgery (Third Faculty of Medicine, Charles University, Prague, Czech Republic). HDF cultures were expanded in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS; Biochrom, Berlin, Germany) and penicillin/streptomycin (Biochrom). They were further culturing at 37 °C and 5% CO 2 . Cells at passages 7-8 were used in all experiments.
The HaCaT cell line 42 , used in the present study was obtained from Cell Lines Service (Eppelheim, Germany). Cells were cultured in Dulbecco's Modified Eagles Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and antibiotics (streptomycin and penicillin) (all from Biochrom, Berlin, Germany).
In all in vitro experiments, TGF-β1 (PeproTech, London, UK) at 30 ng/mL 43 was used as a positive control to evaluate the ability of HDFs to undergo myofibroblast differentiation. We examined whether LPPO modulates TGF-β1 signaling (both canonical and non-canonical). In parallel, LPPO and/or TGF-β1 effect was also studied in HaCaT cells. A standard cultivation medium was used as control. www.nature.com/scientificreports/ MTS assay. Cell viability and proliferation were determined using colorimetric microculture assay with MTS (3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium) dye (Promega, Madison, WI, USA) 44 . Cells were seeded (HDFs-3000 cells/well; HaCaT-10,000 cells/well) into 96-well-plates in culture medium (10% FBS). After 24 h the medium was replaced with a medium containing LPPO (0.1, 1, 5, 10, 25, 50, and 100 mg/L) in the presence or absence of TGF-β1 (PeproTech). Following the next 72 h MTS (at a final concentration of 0.21 mg/mL) was added into the culture medium. After an additional 3 h, cell proliferation was evaluated by measuring the absorbance at wavelength 490 nm using the automated Cytation™ 3 Cell Imaging Multi-Mode Reader (Biotek, Winooski, VT, USA). The absorbance of control wells was taken as 100%, and the results were expressed as a percentage of the untreated control. The experiments were performed in technical triplicates and repeated three times.
Migration (wound healing) assay. The migration of HaCaT cells was evaluated using a scratch assay.
Briefly, a confluent layer of cells cultured on a 6-well plate was scratched using a pipette tip creating a "wound". Afterwards, the medium was replaced with a medium containing LPPO at 1, 5 and 10 mg/L in the presence or absence of TGF-β1 (PeproTech). The wounded area was photographed at the beginning (0 h) and in 24 h. The migration distance (gap area) was determined using NIS Elements (Nikon, Tokyo, Japan) and expressed as a percentage of the original gap created at 0 h. The experiments were performed in technical duplicates and repeated twice.
Western blot analysis. Based on the MTS assay we selected only non-toxic concentrations (1, 5, 10 mg/L) of LPPO for further western blot analysis. HDFs and HaCaTs were seeded on Petri dishes at the density of 5000 and 10,000 cells/cm 2 , respectively and cultivated for 7 days with LPPO in the presence or absence of TGF-β1. The set of primary and secondary antibodies applied in the analysis is shown in Table S5. The analysis was performed as described previously 45 . Briefly, cells were washed with cold PBS and collected in Laemmli sample buffer (100 mM TRIS-HCL pH approx. 6.8, 10% glycerol, 2% SDS) containing protease and phosphatase inhibitors (Sigma-Aldrich, St. Louis, MO, USA). Immediately after collection, cells were disrupted using a sonicator (QSonica, 40% amplitude, 15 s). After boiling (95 °C, 5 min), samples were separated in SDS-PAGE gel (10% Bis-Tris) and transferred to PVDF membrane using iBlot 2 (Thermo Fischer Scientific). Following 1 h of blocking in 5% NFDM/BSA (non-fat dry milk/bovine serum albumin) dissolved in TBS (tris-buffered saline) with 0.1% Tween at room temperature, membranes were incubated with the primary antibody at 4 °C overnight. The next day, the membranes were washed in TBS-Tween (3 × 5 min) and incubated with HRP-conjugated secondary antibodies for 1 h at room temperature. After incubation, the membranes were again washed in TBS-Tween (3 × 5 min), and protein bands were detected using ECL (SuperSignal West Pico PLUS chemiluminescent Substrate, Thermo Fischer Scientific) and the signal was acquired at MF-ChemiBis 2.0 (DNR Bio-Imaging Systems). β-actin was used as a sample loading control. The chemiluminescent signal of all detected proteins was quantified using the Image Studio (LI-COR) western blot densitometry software and normalized to β-actin.

Imunofluorescence of cultured cells.
HDFs and HaCaTs were seeded at a density of 5000 and 10,000 cells/cm 2 , respectively. Both cells were cultivated with LPPO (1, 5 and 10 mg/L) in the presence or absence of TGF-β1 (PeproTech) for 7 days. The set of primary and secondary antibodies applied in the analysis is shown in Table S6. The analysis was performed as described previously 46 . Briefly, tested specimens were fixed with 2% buffered paraformaldehyde (pH = 7.2) for 5 min. and washed with PBS. Cells were permeabilized by exposure to Triton X-100 (Sigma-Aldrich), and sites for antigen-independent binding of antibodies were blocked by incubation with porcine serum albumin. Commercial antibodies were used at concentrations recommended by the suppliers. All specimens were mounted to Vectashield (Vector Laboratories, Eching, Germany) and inspected by using an Eclipse 90i microscope equipped with filter blocks for the three types of dyes (Nikon, Tokyo, Japan) as well as a Hamamatsu CCD camera (Hamamatsu, Shizuoka, Japan) and a computer-assisted image analysis system NIS (Nikon, Tokyo, Japan).
Animal study. The  The allocation of mice (n = 54) in treatment groups is shown in Table 2. All mice were sacrificed under general anesthesia by cervical dislocation 7 days after surgery. Wound samples were removed for histological, bacteriological culture-based and bacteriological quantitative PCR-based evaluations (see below). In addition, plasma and liver were also examined for the presence of LPPO residues. Bacterium and inoculum preparation. Staphylococcus aureus strain CNCTC 5480 (= ATCC 29213) obtained from the Czech National Collection of Type Cultures (https:// www. szu. cz/ cnctc) was used in the present study. To prepare an inoculum for experimental wound infection, 15 mL of Luria-Bertani (LB) medium (Oxoid Ltd, Basingstoke, UK) in a 100-mL Erlenmeyer flask was inoculated with 50 μL of cell suspension with a density of ≈ 10 8 colony-forming units [CFU] per mL, which was prepared in saline from an overnight CNCTC 5480 culture on a sheep blood agar (BA) plate (Oxoid). The inoculated medium was then cultured at 37 °C with shaking in a thermostatically controlled water bath. After 3.5 h, 1 mL of the mid-exponential growth phase culture was taken and the harvested cells were washed twice and finally diluted in PBS (pH = 7.2). The bacterial quantity was determined by the measurement of the optical density (OD) and based on the known correlation between OD and CFU values the final suspension was adjusted to contain ≈ 2.5 × 10 4 CFU/mL of PBS. The bacterial density of the final suspension was checked by CFU counting using serial dilution and plating onto LB agar plates.
Histology of open wounds. Skin-wound specimens were removed from sacrificed mice and routinely processed for histological staining (fixation in 4% buffered formaldehyde, dehydration using increasing concentrations of ethanol, paraffin embedding, sectioning, and staining). Deparaffinized Sections (2-3 µm thick) were stained with hematoxylin-eosin (H + E-basic staining), Sirius Red (SR-collagen staining) and Gram staining (to distinguish between gram-positive/negative bacteria).