Identification of bacterial biofilm and the Staphylococcus aureus derived protease, staphopain, on the skin surface of patients with atopic dermatitis

Atopic dermatitis (AD) is a chronic inflammatory skin disease characterized by an impaired epidermal barrier, dysregulation of innate and adaptive immunity, and a high susceptibility to bacterial colonization and infection. In the present study, bacterial biofilm was visualized by electron microscopy at the surface of AD skin. Correspondingly, Staphylococcus aureus (S. aureus) isolates from lesional skin of patients with AD, produced a substantial amount of biofilm in vitro. S. aureus biofilms showed less susceptibility to killing by the antimicrobial peptide LL-37 when compared with results obtained using planktonic cells. Confocal microscopy analysis showed that LL-37 binds to the S. aureus biofilms. Immuno-gold staining of S. aureus biofilm of AD skin detected the S. aureus derived protease staphopain adjacent to the bacteria. In vitro, staphopain B degraded LL-37 into shorter peptide fragments. Further, LL-37 significantly inhibited S. aureus biofilm formation, but no such effects were observed for the degradation products. The data presented here provide novel information on staphopains present in S. aureus biofilms in vivo, and illustrate the complex interplay between biofilm and LL-37 in skin of AD patients, possibly leading to a disturbed host defense, which facilitates bacterial persistence.


Results
Visualization of bacterial biofilm in lesional skin of patients with AD. Scanning electron microscopy (SEM) analysis of skin biopsies from S. aureus colonized lesional skin of AD patients demonstrated the presence of bacteria and bacterial biofilm at the skin surface. The results showed biofilm at structures identified in the stratum corneum composed of corneocytes, intricate extracellular matrix material, and bacteria. Moreover, it was also possible to visualize individual bacteria surrounded by extracellular matrix material between the corneocytes (Fig. 1a-c). When biopsies of non-lesional skin derived from a non-infected AD patient ( Supplementary  Fig. S1a,b) and from lesional skin of three AD patients (with S. aureus verified skin colonization) (Supplementary Fig. S1c-h) were investigated using SEM, the results showed that lesional AD skin contained considerably more fibrin, extracellular material, and bacteria when compared to non-lesional skin. Next, an ex vivo model was used to quantify the presence of strongly adherent bacteria, representing bacterial biofilm in skin biopsies from patients with AD. The quantification of adherent bacteria was done by enumerating colony-forming units (CFU) per square centimetre of skin released after the biopsies were washed, vortexed and sonicated. The results showed that both weakly (median value 1.01 × 10 4 CFU/cm 2 ) and strongly (median value 1.38 × 10 4 CFU/cm 2 ) attached bacteria were present in the AD skin samples (Supplementary Fig. S1i).
Biofilm production by S. aureus isolates derived from skin of patients with AD. Biofilm formation was measured among 32 isolates of S. aureus derived from skin of AD patients and molecular typing was performed using ADSRRS-fingerprinting analysis. The results showed that 12 out of 32 of the isolates (38%) have the ability to produce a substantial amount of biofilm in vitro, as detected using the crystal violet method (Supplementary Table S1 and Supplementary Fig. S2). According to the molecular typing data, the capacity to produce biofilm was not associated with any specific strain among the skin-derived S. aureus isolates, (Supplementary Table S1). Notably, six of the isolates showed strong capacity to produce biofilm in vitro, yielding an OD 600 -value in the biofilm assay of more than 0.6 ( Supplementary Fig. S2).
It has been recently shown by SEM that the organization of S. aureus bacteria undergo gradual morphological changes at different growth phases during biofilm formation 22 . To determine if the differences in biofilm formation were dependent on the growth of the bacterial isolates, the optical density of liquid cultures was measured spectrophotometrically. The results showed that growth rates were approximately similar among the tested S. aureus strains ( Supplementary Fig. S3), and no significant differences between strong and weak biofilm-producing strains were observed.

Antimicrobial effects of LL-37.
To test if the human cathelicidin peptide LL-37 exerted different antimicrobial effects on planktonic bacteria and on bacterial biofilm, a modified version of the minimal biofilm eradication concentration (MBEC) assay was used on various biofilm producing S. aureus isolates, all derived from skin of patients with AD. The results showed a marked difference between the minimum inhibitory concentration (MIC) values and the MBEC values, and indicated that the concentration of LL-37 needed to eradicate bacterial biofilms exceeded the highest tested concentration of 160 μM (Table 1). In contrast, the MIC values obtained for planktonic cells were in the range of 10-20 μM LL-37 (Table 1).

Visualization by confocal light scanning microscopy (CLSM) of human LL-37 binding to bacterial biofilm.
To determine whether LL-37 binds to the bacterial biofilm, mature (6-day old) biofilms produced by three S. aureus isolates, derived from lesional skin of three AD patients, were grown on Cellview TM cell culture dishes with glass bottom. The bacterial cells in the biofilm were stained with the fluorescent stain SYTO ® 9 green and incubated with TAMRA labeled LL-37. The results showed that the LL-37 peptide bound to the bacterial cells in the biofilms (Fig. 2). When the Z-stacks were analyzed, and quantification of the intensity obtained by Profile plug-in (Zeiss ZEN Confocal Software), results indicated three distinctly different populations of fluorophores in the samples; red (TAMRA labeled LL-37) and green (SYTO ® 9) seen separately, and yellow representing co-existence of bacterial cells and the TAMRA-labeled LL-37 peptide. Binding of TAMRA labeled LL-37 to the S. aureus cells was observed throughout the biofilm (Supplementary Fig. S4a-c).

S. aureus staphopain and effects on LL-37.
Previous investigations demonstrated that LL-37 could be degraded by the staphylococcal proteases aureolysin and V8 protease 19 . Here, we explored whether the S. aureus derived protease staphopain was produced by the bacteria in vivo and if it could cleave LL-37 in vitro. First, biopsies from lesional skin of patients with AD were examined using electron microscopy, with gold-labeled antibodies against the protease. The results showed that the S. aureus staphopain was indeed present adjacent to coccoid bacteria in the biofilm (Fig. 3a,b). As SspB has been particularly associated with virulence 20 , we first investigated if this protease was able to degrade LL-37. As demonstrated by gel-electrophoresis, LL-37 was degraded into fragments of lower molecular weight by the enzyme. The digested material was analyzed by mass spectrometry (MS) and the results revealed several low molecular weight peptide fragments (Fig. 3c,d). Moreover, when LL-37 was subjected to ScpA the results did not reveal any obvious major cleavage of LL-37 as demonstrated by gel-electrophoresis, although the MS data detected cleavage of minor fragments ( Supplementary Fig. S5). These data were thus in correspondence with previous findings of SspB as a major virulence factor.
To investigate the antibacterial effects of the peptide fragments on planktonic bacteria, a radial diffusion assay was performed. The results demonstrated that only one fragment, FKR21 was antimicrobial against the Gram-positive S. aureus and Gram-negative E. coli bacteria (Fig. 4a).
Fragments of LL-37 have previously been ascribed both pro-and anti-inflammatory effects [23][24][25][26][27][28][29] . Table 2 compares the identified fragments with previously published LL-37 derived peptide sequences, and references are included illustrating reported immune-modulating actions. To explore if these staphopain-generated peptide fragments also showed immune-modulatory effects in vitro, four peptide sequences from the C-terminal part of LL-37 identified by MSMS were synthesized (FKR10, FLR11, LLG11 and FKR21) ( Table 2 and Fig. 3d). The peptides were then added to lipopolycaccharide (LPS)-or lipoteichoic acid (LTA)-stimulated human monocytic cells (THP1-XBlue-CD14) to address effects of the fragments on activation of the transcription factors NF-κB and   AP-1, central initiators of pro-inflammatory cytokine production 30,31 . The results showed that that only FKR21 dose-dependently reduced LPS-and LTA-induced NF-κB and AP-1 activation (Fig. 4b), the results of LL-37 are included for comparison (Fig. 4c). To investigate the impact on viability of keratinocytes, by the LL-37 derived peptide fragments, HaCaT cells were subjected to the fragments. The viability was analysed by utilizing the MTT assay. As shown, the HaCaT cells were not significantly affected by the LL-37 derived peptide fragments (Fig. 4d).
It is of note that the analysis also revealed an additional minor fragment (LLG16) (Fig. 3d, lower panel), which, interestingly, corresponds to similar fragments previously characterized 23 and generated by processing of cathelicidin by skin proteases 32 .
LL-37 effects on biofilm production. Although we could not find any substantial effect of LL-37 on eradication of biofilm assessed by the Calgary Biofilm Device (CBD)-method, LL-37 might still have inhibitory effects on biofilm formation. To investigate this, an abiotic solid surface assay was performed as described by Hell et al. 33 , with minor modifications. The results showed that LL-37 concentrations above 20 μM significantly inhibited biofilm formation of S. aureus isolates being high producers of biofilm (P < 0.05, Wilcoxon rank sum test), whereas no or little inhibitory effect was observed on low biofilm producing isolates, as well as the ATCC29213 strain ( Supplementary Fig. S6a). The LL-37 derived fragments yielded no effects on biofilm formation at the same doses as used for LL-37 ( Supplementary Fig. S6b). Moreover, S. aureus isolates were subjected to LL37 and FKR21 in low salt and physiologic salt conditions. The results showed that LL-37 was less inhibited by salt when compared to the results obtained with the peptide FKR21 ( Supplementary Fig. S7).

Discussion
The main findings in this study are the identification of bacterial biofilm in lesions of S. aureus colonized AD skin combined with findings that S. aureus isolates of patients with AD are able to produce a substantial amount of biofilm, which in turn, may protect S. aureus from LL-37-mediated killing. Since biofilms are typical for later stage growth phases, our data correspond well with results showing that protease production of S. aureus reaches maximal activity in the post-exponential growth phase 20 . Our data also correspond well with recent reports identifying biofilms and glycocalyx structures on skin from AD patients, along with biofilm producing S. aureus isolates 7,9,34 . Taken together, all these observations, combined with the fact that staphopains are among the most copiously produced proteases of S. aureus 20,35 , clearly motivate further studies on the clinical importance of biofilms for bacterial persistence and protease activity in patients with AD. LL-37 is one of the most well characterized AMPs found in human skin particularly under inflammatory conditions, present in the specific granules of neutrophils and in keratinocytes [36][37][38] . The level and expression of LL-37 is reported to range from ≈1 μM in human sweat, to higher local concentrations in inflamed skin, such as in acne rosacea 17,18 . S. aureus biofilms are composed of, not only bacteria, but also an extracellular matrix comprising multiple macromolecules, including polysaccharides (such as polysaccharide intercellular adhesin), extracellular DNA and proteins 39 . It is therefore likely that interactions between biofilm substances such as negatively charged polysaccharides and LL-37 can lead to scavenging and inactivation of the peptide´s antimicrobial activity. Compatible with these observations are the findings with confocal microscopy, showing binding of LL-37 to the in vitro grown biofilm. It is also of note that at high concentrations, LL-37 showed inhibitory effects on biofilm formation of S. aureus. Hence, our results correspond with earlier reports on LL-37 and biofilm formation, however, the effectiveness of LL-37 in disrupting S. aureus biofilm seems to be less in comparison to other peptides 40,41 .  Although the exact mechanisms of how LL-37 performs its anti-biofilm effects are not fully elucidated, the central fragment of the peptide seems to be important for its anti-biofilm effects 42 . The findings that S. aureus proteases from the biofilm, such as staphopains, were able to degrade LL-37, generating peptide fragments with modified or abrogated antibacterial effects on planktonic bacteria as well as bacterial biofilms, may have important implications. It has been previously reported that peptide fragments produced by degradation of LL-37 can exert pro-inflammatory actions in skin 18 . The degradation of LL-37 by staphopains may therefore yield comparable pro-inflammatory effects which in turn, might further amplify the inflammatory process and the "pathogenic vicious loop" typical for AD 43 . In our in vitro studies one of the low molecular weight peptide fragments produced by degradation of LL-37 (FKR21) was able to reduce TLR4-and TLR2-mediated responses, indicating that some fragments may have retained anti-inflammatory actions. Interestingly, a fragment produced by the staphopain SspB (LLG16), is almost identical to a fragment generated by processing of cathelicidin by skin proteases (LL-17) 32 . This suggests a functional overlap between the activities of endogenous proteases such as kallikreins, and bacterial enzymes such as SspB. Furthermore, bacterial proteases found in biofilms can also provide a reservoir of enzymes, that directly activate proenzymes, such as kallikreins, or perhaps cleave structural proteins of importance for maintaining a permeability barrier 44,45 .
As mentioned in the Introduction, staphylococci may influence various pathomechanisms in AD. The results by Allen et al. 7 indicate that staphylococci may activate toll like receptor 2 (TLR2) 7 and up-regulate the expression and production of several pro-inflammatory cytokines 14 . Even though it is reported that patients with AD show less response to TLR-2 stimulation than healthy controls 46 , several products from S. aureus are known to induce a pro-inflammatory response in keratinocytes 47 . Furthermore, superantigens from staphyloccoci may directly stimulate T-lymphocytes via the T-cell receptor 48 . It is also reported that protease production by S. aureus causes skin barrier dysfunction 49 and that S. aureus colonization is associated with impairments of the skin barrier in AD 4 . Chronic biofilm infections by S. aureus can activate the immune system generating damage of host tissue, and therefore generate an environment facilitating the persistence of the biofilm 50 . Moreover, suppression of the inflammatory response seems to prevent the development of chronic biofilm infections 51 .
Biofilm production by staphylococci could constitute an overlooked pathogenic mechanism, leading to inactivation of AMP-mediated bacterial killing. It is also notable that S. aureus derived proteases, such as staphopains (SspB and ScpA) are known to inhibit biofilm formation, and the addition of ScpA to bacterial biofilm was shown to be able to disperse an established biofilm 52 . Thus, under conditions with up-regulated staphopain production, these proteases may not only degrade endogenous AMPs, but also counteract biofilm formation and disperse the S. aureus biofilm. Whether this enables bacterial spread and skin infection under certain circumstances remains to be investigated.
In conclusion, our results demonstrate the existence of S. aureus biofilms and staphopains in AD skin. The biofilm protects the bacteria from LL-37 mediated killing, and further, staphopains may cleave LL-37 into smaller fragments. Biofilm formation by S. aureus strains could thus support a persistent bacterial colonization in AD skin. Moreover, S. aureus biofilms could be a source of proteolytic enzymes at the skin surface, with the capacity to cleave endogenous AMPs and interfering with the epidermal inflammatory response. Patients and skin biopsy. Adult patients aged 18 years or over, with AD verified by the UK refinement of the Hanifin and Rajka diagnostic criteria for AD 53,54 , were recruited from the Dermatology Clinic at Lund University Hospital, Lund, Sweden. Bacterial samples were taken from the skin of the patients and tissue biopsies from AD lesional areas were processed as per standard procedure, for description see section electron microscopy and immuno-staining. The participants gave informed consent complying with the Helsinki Declaration, all methods and experiments were performed in accordance with relevant guidelines and regulations, and the Regional Ethics Examination Board of Lund, Sweden approved the study (Permit Numbers: 144/2010, 317/2010, 82/2012).

Bacterial isolates. Staphylococcus aureus ATCC 29213 isolate was from the American Type Culture
Collection (Rockville, MD, USA). Clinical bacterial flora specimens were obtained from the skin of AD patients using a modified version of a method described by Williamson and Kligman 55 . Briefly, a sterile plastic cylinder (3.8 sq cm) was applied to the area of the skin to be scrubbed. Then, a sterile solution of 1 ml 0.05% Triton X-100 in 0.075 M phosphate buffer was pipetted into the plastic cylinder and the area scrubbed for 60 seconds with a sterile Servant ® disposable inoculation loop and needle (Konstrumed oy, Tempere, Finland). For identification of bacteria, the samples were processed at the Department of Clinical Microbiology at Skåne University Hospital in Lund, Sweden, following standard routines for identification of S. aureus. Molecular typing was performed using ADSRRS-fingerprinting analysis as previously described 56,57 . Briefly, DNA from S. aureus was digested with the two restriction enzymes BamHI (10 U/μl) (Sigma) and Xbal (10 U/μl) (Sigma). Cohesive ends of DNA were ligated with adapters and then amplified. The PCR products were electrophoresed on polyacrylamide gels and then stained with ethidium bromide. The gels were photographed under UV-light. The strains used in this study represent a variety of different genotypes (Supplementary Table S1).
Biofilm assay. Quantification of biofilm formation of S. aureus isolates was performed in a 96-well microtiter plate assay as previously described, with minor modifications 58 (See Supplementary Methods S1).

Minimum inhibitory concentration (MIC) and minimal biofilm eradication concentration (MBEC).
Measurement of the bacterial susceptibility was determined by using a modified version of the CBD method (MBEC TM Biofilm Inoculator, Innovotech, Edmonton, Canada) 59 following the manufacturers protocol. Briefly, to establish biofilms in vitro using the CBD, 150 μl bacterial solution of 10 5 CFU/ml in 1.5% tryptic soy broth (TSB) was supplemented with 0.3% glucose and aliquoted into each well of the MBEC plate and thereafter incubated for 24 h in a rotary incubator at 37 °C and 180 rpm. This allowed the biofilm to form on the pegs. The pegs with established biofilms were then washed once in sterile PBS (200 μl/well). The LL-37 peptide was serially diluted in Müller Hinton broth (MHB) in a 96-well polypropylene microplate (Costar ® , Corning, NY, USA) and incubated for 16 h according to the NCSLA guidelines, as described by Wiegand et al. 60 .
The MIC values measured using the CBD are equivalent to MIC values obtained using the National Committee for Clinical Laboratory Standards (NCCLS) procedure 59 . The MIC value, representing the concentration required to inhibit growth of planktonic bacteria, was determined from bacteria that were shed from the pegs when the lid of the CBD was placed in different concentrations of LL-37. Following determination of the MICs for planktonic cells, the MBEC pegs were washed once in sterile PBS and placed into the recovery plate containing 1.5% TSB supplemented with 0.3% glucose and universal neutralizer following the manufacturers protocol (200 μl/well). Radial diffusion assay (RDA). RDA was performed as previously described 61,62
Fluorescent staining and CLSM. Bacterial cell cultures of S. aureus isolates were suspended in 3% TSB and shaken at 180 rpm in an incubator at 37 °C for 18 h. A bacterial solution of 10 7 CFU/ml in 1.5% TSB supplemented with 0.3% glucose was prepared. In vitro biofilms were then established by adding 1 ml aliquots of the bacterial solution to sterile Cellview TM cell culture dishes with glass bottom (Greiner Bio-One, Germany), incubated at 37 °C for indicated time periods, and the medium was changed every second day. The medium was removed and the culture dishes were gently washed using PBS solution. For CLMS, the remaining adherent microbial biofilms present on the bottom of the dishes were stained by adding 200 μl/dish of a dilution of 3 μl SYTO ® 9 green fluorescent stain (Live/dead ® BacLight TM bacterial viability kit L7012, Molecular Probes, USA) into 1 ml distilled water, and incubated at 37 °C for 15 min in the dark. Subsequently, the samples were incubated with 200 μl of 5 μM TAMRA-labeled LL-37 (Innovagen AB, Lund, Sweden), followed by incubation at 37 °C for 15 min in the dark. The samples were then fixed with 2% freshly prepared formaldehyde solution and incubated for 5 min on ice followed by 25 min at room temperature. Subsequently, the material was mounted on glass coverslips using Dako fluorescent mounting medium (S3023, Dako Sweden AB, Stockholm, Sweden). After each step of staining and fixation the biofilm was gently washed twice with PBS. Images were captured using a Zeiss laser-scanning microscope 510 (Carl Zeiss, Jena, Germany). An objective lens (plan aperture, 20x magnification) was used. The image stacks collected by CLSM were analyzed or processed with Zeiss Efficient Navigation (ZEN) 2009 software (Carl Zeiss, Germany) and by the ImageJ software.
Electron microscopy and immuno-staining. Transmission immunoelectron microscopy was performed as described earlier 63 . In short, specimens were fixed in 150 mM sodium cacodylate, 2.5% glutaraldehyde, pH 7.4 and embedded in Epon. After antigen retrieval with sodium metaperiodate, specimens were incubated with primary antibodies, followed by detection with species-specific secondary antibody-gold conjugates. Samples were examined in a Philips/FEI CM 100 TWIN transmission electron microscope (FEI Co, Hillsboro, Oregon, USA) at 60-kV accelerating voltage. Images were recorded with a side-mounted Olympus Veleta camera with a resolution of 2048 × 2048 pixels (2 k × 2 K) using ITEM TM software. For scanning electron microscopy, specimens were fixed over night at RT with 2.5% glutaraldehyde in 150 mM cacodylate, pH 7.4. After washing with cacodylate buffer, they were dehydrated with an ascending ethanol series from 50% (v/v) to absolute ethanol, and subjected to critical point drying with carbon dioxide. Tissue samples were mounted on aluminum holders, sputtered with 20 nm palladium/gold, and examined in a Philips/FEI XL 30 FESEM scanning electron microscope using an Everhart-Tornley secondary electron detector or in DELPHI, Phenom-World. Image processing was done with the Scandium software for simple image acquiring and auto-storage into the Scandium database. All electron microscopic work was performed at the Core Facility for Integrated Microscopy, Panum Institute, University of Copenhagen and at Infection Medicine, Lund University. Contrast, brightness and pseudocolours were adjusted in Adobe Photoshop CS6.
Ex vivo model, quantification of strongly adherent bacteria in lesional AD skin (See Supplementary Methods S1).

Proteolytic degradation of LL-37 and SDS-PAGE analysis (See Supplementary Methods S1).
Mass spectrometric identification of LL-37-derived peptides. Matrix-assisted laser desorption ionization mass spectrometry (MALDI-MS) was used for the identification of trypsin generated peptides of LL-37. A MALDI matrix solution consisting of 5 mg/ml hydroxycinnamic acid in 50% acetonitrile, 0.1% (v/v) phosphoric acid 67 was mixed with the staphopain-generated peptides. The matrix solution contained two peptide standards [des-arg-bradykinin (m/z 904.468) and ACTH 18-39 (m/z 2465.199)] that were used for internal mass calibration in every analyte/matrix position. MALDI-MS and MS/MS analyses of the samples were performed on a 4700 Proteomics Analyzer MALDI-TOF/TOF ™ mass spectrometer (Applied Biosystems, Framingham, MA).
Database searching was carried out using Mascot (Matrix Science) with Swissprot as the database and with a peptide mass tolerance of 50 ppm and a fragment mass tolerance of 0.2 Da.
Statistics. Data are presented as means ± standard deviation of the means. To describe the differences between groups, one-way ANOVA with Dunnet's multiple comparisons test was used. In order to determine significant differences between two groups, the Wilcoxon rank-sum test was used, and p < 0.05 was considered as significant. The statistical software used was GraphPad PRISM ® version 6.0c (GraphPad Software, Inc., La Jolla, CA, USA).
Data Availability. The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.