Proteolytic Degradation of reduced Human Beta Defensin 1 generates a Novel Antibiotic Octapeptide

Microbial resistance against clinical used antibiotics is on the rise. Accordingly, there is a high demand for new innovative antimicrobial strategies. The host-defense peptide human beta-defensin 1 (hBD-1) is produced continuously by epithelial cells and exhibits compelling antimicrobial activity after reduction of its disulphide bridges. Here we report that proteolysis of reduced hBD-1 by gastrointestinal proteases as well as human duodenal secretions produces an eight-amino acid carboxy-terminal fragment. The generated octapeptide retains antibiotic activity, yet with distinct characteristics differing from the full-length peptide. We modified the octapeptide by stabilizing its termini and by using non-natural D-amino acids. The native and modified peptide variants showed antibiotic activity against pathogenic as well as antibiotic-resistant microorganisms, including E. coli, P. aeruginosa and C. albicans. Moreover, in an in vitro C. albicans infection model the tested peptides demonstrated effective amelioration of C. albicans infection without showing cytotoxity on human cells. In summary, protease degradation of hBD-1 provides a yet unknown mechanism to broaden antimicrobial host defense, which could be used to develop defensin-derived therapeutic applications.


Results
Degradation of reduced hBD-1 generates an antimicrobial octapeptide. Reduction of the three disulphide bridges of hBD-1 yields a linear peptide which not only differs structurally from the oxidized form 11 , but is also more prone to proteolytic degradation by the protease trypsin 16 . To further analyze proteolytic susceptibility of hBD-1 (Fig. 1A) towards physiological gastro-intestinal proteases, we treated oxidized and reduced hBD-1 with pepsin and chymotrypsin (Fig. 1B). Similar to trypsin digestion, oxidized hBD-1 was protease resistant while the reduced form was readily digested. Focusing on the degradation products we detected a fragment having an m/z of 893.5, corresponding to the eight carboxy-terminal amino acids of hBD-1, NH 2 -RGKAKCCK-COOH (RGKAKCCK). To assess the in vivo relevance of our findings in more detail, we incubated oxidized and reduced hBD-1 with human duodenal secretion, which is rich in proteolytic enzymes. In agreement with our in vitro data, ex vivo digestion generated a fragment having an m/z of 893.5 (Fig. 1C) for reduced hBD-1 but not for oxidized hBD-1. While the in vivo presence of the octapeptide in the human gut remains to be proven, this finding supports the hypothesis that proteolytic cleavage of reduced hBD-1 could generate a novel antimicrobial peptide in the human intestine. To test whether antibiotic activity is retained in this degradation product we next analyzed antimicrobial activity of the terminal octapeptide. In a radial diffusion assay (RDA) 17 we thus compared activity of oxidized and reduced hBD-1 with the octapeptide RGKAKCCK against selected commensal and pathogenic microorganisms (Fig. 1D). The octapeptide exhibited convincing activity against Bifidobacterium adolescentis, Streptococcus salivarius ssp. thermophilus, Escherichia coli, Candida albicans but neglectable antibiotic activity against Pseudomonas aeruginosa. While RGKAKCCK and oxidized hBD-1 generated no inhibition zones against Bifidobacterium breve, only reduced hBD-1 inhibited its growth. As expected, however, semi-quantitative evaluation on a molar base (4 µg reduced hBD-1 is equivalent to 254 µM in the RDA while 1 µg of octapeptide is equivalent to 280 µM) revealed that activity of the isolated terminus is less potent compared with the full length-peptide (Fig. 1D).
We found previously that cysteine residues are crucial for antimicrobial activity of hBD-1 against E. coli and Bif. adolescentis 11,12 . We confirmed these results for the terminal octapeptide, as replacing both cysteines (RGKAKAAK) completely abolished activity against E. coli and C. albicans (Fig. 1E, Supplementary Fig. 1). However, replacing either Cys 6 (RGKAKACK) or Cys 7 (RGKAKCAK) by alanine resulted in strongly decreased activity against C. albicans, while antibiotic activity of those peptides increased against E. coli. Consequently, cysteine residues seem to have an important, yet different, role for the antibiotic mechanism against the tested fungi and bacteria.
As antimicrobial peptide activity also relies on a positive net charge 18,19 we further investigated the role of positively charged amino acids (Fig. 1E). Despite having two cysteine residues, the variants lacking Arg 1 (AGKAKCCK, net charge +3) or Arg 1 and Lys 3 (AGAAKCCK, net charge +2) were completely inactive against both tested microorganism. Thus, antimicrobial activity of the octapeptide RGKAKCCK (net charge +4) depends on cysteine residues as well as a particular positive charge. Yet, as a reversed-order peptide had lower activity than the wild-type peptide or a scrambled version, especially against C. albicans (Fig. 1F), not only the amino acid composition but also its sequential order or its position seem to be involved in the peptides' activity.
Characterization of RGKAKCCK and its modified variants. Since the discovery of antimicrobial peptides there is anticipation to exploit them as antibiotic drugs 20 . To test the potential of our octapeptide to be used as a therapeutic agent, we first generated peptide variants to improve its stability. To prevent non-specific cleavage by amino-carboxypeptidases, we chemically stabilized its termini by amino-terminal acetylation and carboxy-terminal amidation (Ac-RGKAKCCK-NH 2 ) and generated both peptides also in D-amino acid configuration (rGkakcck and Ac-rGkakcck-NH 2 , respectively).
Next, to evaluate the antibiotic activity, we tested wild-type and modified peptide variants in their ability to inhibit growth of (opportunistic) pathogenic microorganisms ( Fig. 2A). Direct comparison revealed that those variants with stabilized termini had promising activity against E. coli and C. albicans. Moreover, while all peptides displayed antimicrobial activity at pH 7.4, a reducing environment (DTT), acidification (pH 5.7), or a combination of acidification and reducing conditions (pH 5.7 + DTT) strongly decreased antimicrobial activity against E. coli (Supplementrary Fig. 2a) and C. albicans (Supplementrary Fig. 2b). Thus, antimicrobial acitivity of the generated octapeptides can be influenced by environmental conditions, in particular by reducing conditions with an acidic pH.
The widespread use of antibiotics in agriculture and to treat bacterial infections has led to a rapid emergence of microbial resistance 21,22 . As a consequence, in hospitals several multi-drug resistant strains exist that threaten effective therapy of microbial infections 23 . We therefore tested if the hBD1-derived peptide and its modified forms are also active against drug-resistant clinical isolates. As shown in Fig. 2B, we identified antimicrobial activity against clinical isolates of antibiotic-resistant E. coli, Pseudomonas aeruginosa and Enterococcus faecium. In contrast, Acinetobacter baumanii was only susceptible towards Ac-rGkakcck-NH 2 whereas Enterococcus faecalis and K. pneumoniae were not sensitive. Direct comparison of the tested peptide revealed that the peptides RGKAKCCK and Ac-rGkakcck-NH 2 inhibited growth of the most tested antibiotic-resistant bacteria, making them the most promising candidates for further drug development among the four tested peptides.
As the radial diffusion assay does not differentiate between microbistatic and microbicidal activity and contains immobilized bacteria or fungal cells, we complemented our antimicrobial tests with a broth microdilution assay to investigate susceptibility of E. coli and C. albicans. The tested octapeptides completely inhibited C. albicans growth at concentrations of 100 µg/ml, while the growth of E. coli was only inhibited to 40% of the untreated www.nature.com/scientificreports www.nature.com/scientificreports/ control by Ac-rGkakcck-NH 2 (Fig. 2C). In addition, microbial cultures were incubated with the different peptides and colony forming units (CFU) were determined. For E. coli incubated with 100 µg/ml of RGKAKCCK or Ac-RGKAKCCK-NH 2, a more than 100-fold decrease in CFU was observed when compared to untreated controls (Fig. 2D). In contrast, CFU reduction of C. albicans was less pronounced, thus, indicating bactericidal activity against E. coli and a combination of fungicidal and fungistatic activity against C. albicans of the tested octapeptides.
Many antimicrobial peptides target the microbial membrane 24,25 . To test whether this is also true for the octapeptides, we used a flow cytometric assay measuring membrane permeability by the dye propidium iodide (PI), which cannot permeate intact membranes. E. coli and C. albicans were incubated with the peptides and PI uptake, Figure 2. Antimicrobial activity of RGKAKCCK and its modified variants. The carboxyterminal octapeptide RGKAKCCK was stabilized at its termini by acetylation of the amino-terminus and amidation of the carboxyterminus (Ac-RGKAKCCK-NH 2 ). Both variants were also synthesized by using D-stereoisomeric amino acids (indicated by small letters). Antimicrobial activity of octapeptides (4 µg) was tested against pathogenic microorganisms of clinical relevance (A) and with antibiotic resistance (B) in radial diffusion assay. 3-MRGN: multi-resistant Gram negative pathogen (3 out of 4 antibiotic classes), 4-MRGN: multi-resistant Gram negative pathogen (4 out of 4 antibiotic classes) MRSA: Methicillin-resistant Staphylococcus aureus, Amp/Imi: Ampicillin/Imipinem, Van: Vancomycin. (C) Different concentrations of octapeptides were tested in a turbidity liquid assay against E. coli ATCC 25922 and C. albicans ATCC 10231. Peptides were incubated with tested microorganisms and change in optical density (OD 600nm ) was measured and % growth of untreated control was plotted after 12 hours. (D) Aliquots were plated on agar plates and colony forming units (CFUs) were calculated the next day. Data are presented as mean +/− SEM of at least three independent experiments. The statistical significance was evaluated by using Kruskal-Wallis test compared to control and marked with *p < 0.05 and **p < 0.01. (2019) 9:3640 | https://doi.org/10.1038/s41598-019-40216-2 www.nature.com/scientificreports www.nature.com/scientificreports/ which indicated cell death, was analysed. For E. coli, all tested peptides exhibited at least 70% bacterial killing, while treatment with RGKAKCCK, Ac-RGKAKCCK-NH 2 or Ac-rGkakcck-NH 2 led to almost 100% cell death (Fig. 3A). In contrast, only the wild-type peptide RGKAKCCK induced up to 60% PI uptake when incubated with C. albicans.
In addition to PI uptake, we analyzed cellular membrane potential by using the membrane potential sensitive dye DiBAC 4 (3) (Fig. 3B). When incubating the peptides with E. coli, we observed strong membrane depolarization for the same peptides that caused PI uptake. In contrast, C. albicans displayed less than 20% of membrane depolarization. Thus, our results support a bactericidal effect of RGKAKCCK and Ac-rGkakcck-NH 2 against E. coli by targeting the bacterial membrane, while the antibiotic effect against C. albicans seems rather membrane-independent.
To further investigate whether octapeptide treatment leads to structural damage of the microorganisms we used transmission electron microscopy (TEM) to visualize peptide-treated bacteria and fungi. Incubation of E. coli and C. albicans with all tested variants of the octapeptide led to different degrees of structural disintegration (Fig. 3C). This was especially pronounced in E. coli, where all peptides caused detachment of the cell membrane from the cytosol, cell wall and membrane disruption as well as disintegration of cytosolic structures. In C. albicans, however, no destruction of the fungal cell wall could be observed, confirming results from the flow cytometric expriments (Fig. 3A,B). Still, disintegration of cytosolic structures could also be observed in fungal cells, similar to the damage observed in E. coli. Taken together, our results demonstrate a bactericidal and fungicidal/fungistatic effect of the tested octapeptides. potential of the octapeptides for therapeutic drug development. For potential therapeutic application, toxicity of the peptides needs to be excluded. We therefore analyzed cell-toxicity of all peptide variants against the intestinal epithelial cell line CaCo-2 and against erythrocytes. By utilization of a WST viability test we could not observe any cell toxicity in the relevant antimicrobial concentration ranges of 100 µg/ml (Fig. 4A) and 200 µg/ml (Supplementrary Fig. 4a) or any hemolytic effect against erythrocytes ( Fig. 4B and Supplementrary Fig. 4b). www.nature.com/scientificreports www.nature.com/scientificreports/ We furthermore tested the toxicity of the octapeptides by using in vitro reconstituted human oral epithelium (RHOE), which was analyzed by histology and lactatdehydrogenase (LDH) activity (Fig. 4C). No LDH release was detected with peptide variants, whereas the cytotoxic compound Triton-X-100, used as a positive control, induced a strong release of LDH. Similarly, when quantifying the release of the pro-inflammatory cytokines Interleukin-8 (IL-8) and IL-1α after incubation of the RHOE with the octapeptides we did not detect any inflammatory response in the RHOE (Fig. 4D).
Suitability of the octapeptides as novel candidates for antimicrobial drug development was further substantiated in a model or oral candidiasis 26 . In this model a multilayer of RHOE was preincubated with 50 µg/ml (Fig. 5A) or 100 µg/ml octapeptides (Supplementrary Fig. 5) for 1 h before infecting the cells with C. albicans for 24 h (Fig. 5A). Epithelial damage of RHOEs was quantified by independent experts in a blinded manner on a scale between 0 and 5 (Fig. 5B). Untreated cells infected with C. albicans displayed evident epithelial damage www.nature.com/scientificreports www.nature.com/scientificreports/ and cell lysis and detectable fungi (coloured in red) in all layers of the epithelium. However, pre-treatment with RGKAKCCK convincingly reduced the fungal load in the epithelium and ameliorated epithelial damage. Additionally, a histological analysis confirmed the protective effect of RGKAKCCK (Fig. 5B) whereas a pre-treatment with 50 µg/ml rGkakcck moderately improved epithelial damage. In contrast, epithelial damage and a high amount of detectable fungal cells in lower epithelium were observed with a pre-treatment with the modified octapeptides Ac-RGKAKCCK-NH 2 and Ac-rGkakcck-NH 2 : While C. albicans cells invaded the whole epithelium with a pre-treatment with Ac-RGKAKCCK-NH 2 , C. albicans invaded and damaged only the upper epithelium with a pre-treatment with Ac-rGkakcck-NH 2. Consequently, our eight amino-acid peptide RGKAKCCK of the carboxy-terminus of hBD1 has the best potential to be optimized for topical application against infectious microbes on epithelial surfaces.

Discussion
So far, no multicellular organism has been identified which does not produce antimicrobial peptides. Even over a long time of evolution those host defense molecules have retained their antimicrobial capacity with only minor resistance mechanisms at the microbial target 15,27 . Thus, an improved knowledge of these antibacterial molecules may help to identify novel targets for antimicrobial therapy 28 . In fact, in the recent years short antimicrobial peptides have gained increased interest as new opportunities for therapeutics 29 . In here, we characterized a carboxy-terminal octapeptide of hBD-1 that was generated after proteolytic digestion by gastro-intestinal proteases and might thus occur in vivo in the human gastrointestinal tract. Subsequently cells were infected with C. albicans SC5314 (highlighted in red). Representative images are shown (magnification 400x, bar = 100 µm). (B) Epithelial damage was evaluated by four independent experts and the combined evaluation (mean +/− SEM, criteria described in methods) is shown. The statistical significance was evaluated by using Kruskal-Wallis test with ns = not significant and *p < 0.05. www.nature.com/scientificreports www.nature.com/scientificreports/ While reduction of disulphide bridges increases activity of several antimicrobial peptides 11,30,31 , it also increases their susceptibility towards proteolytic digestion. This is especially relevant in the intestinal tract, where duodenal secretions contain high amounts of proteases to facilitate digestion of dietary proteins. However, our observation that a degradation product of hBD-1 retains antibiotic activity indicates effective utilization of a scare resource: we speculate that once the reduced peptides diffuse from the intestinal mucus layer towards the lumen, they can be degraded as soon as they come into contact with intestinal proteases. But instead of being inactivated, degradation products can retain their antimicrobial activity, thereby making the most effective use of these antimicrobial host defense molecules.
While a major shift in antimicrobial activity and activity spectrum can be observed after reduction of the oxidized hBD-1 (Fig. 1D, consistent with 11,12,32 , only minor alterations can be observed after degradation of the reduced peptide. However, on a molar basis the octapeptide does not exhibit the full potency of the full-length peptide, suggesting that the remaining 28 amino acids of hBD-1 hold additional features that enhance and modulate antibiotic activity as, for instance, observed for Bifidobacterium breve (Fig. 1D). Remarkably, and in contrast to hBD-1, a reducing environment diminished antimicrobial activity of the tested octapeptides against E. coli and C. albicans (Supplementrary Fig. 2). It is possible that due to their small size the peptides need to form dimers or higher-order oligomers to effectively penetrate the microbial cell wall. It is likely that such oligomerization occurs by forming intermolecular disulphide-bridges, thus explaining the strong dependence on cysteine residues for the activity of RGKAKCCK against E. coli and C. albicans (Fig. 1E). We could observe that human blood serum can influence bacterial growth and the antimicrobial activity ( Supplementrary Fig. 3). Similarly, acidic pH led to lower antimicrobial activity as compared to pH 7.4 (Supplementrary Fig. 2). This is in accordance with previous studies, which could show that antimicrobial peptides can bind to human plasma proteins 33,34 . Thus, further optimization of a potential peptide formulation would be required to employ the octapeptide as a topical skin therapeutic, as human skin has a pH of about 5.5.
The cell envelope is a commonly discussed target for antimicrobial peptides 35 . While the full length hBD-1 targets the bacterial cell wall and entrapped bacteria in net-like structures, the octapeptides seem to have distinct antimicrobial mechanisms (Fig. 2D). Our methods revealed a breakdown of membrane potential and loss of membrane integrity in bacteria. In contrast, treated fungal cells displayed a functional membrane but cytosolic defects (Fig. 3). These data highlight that our octapeptides have diverse antibiotic strategies for different microorganisms. Remarkably, even for the same microbial species, we observed strain specific differences in susceptibility. While the P. aeruginosa ATCC type strain was not susceptible towards our octapeptides (Figs 1D and 2A), the multi-resistant P. aeruginosa 4-MRGN strain was susceptible (Fig. 2B). This is in accordance with previous studies, which could show that antibiotic-resistant bacteria show an increased sensitivity against antimicrobial peptides 36 .
Different peptides derived from β-defensins have already been investigated on their antimicrobial activity against bacteria and fungi 37,38 . For instance, 19-mer peptides derived from the carboxy-terminus of hBD-1 or −2 and a 22-mer derived from the terminus of hBD-3 were analyzed on their antibiotic activity. These peptides retained one disulphide bridge and had lethal concentrations in the low micromolar range. With our carboxy-terminal octapeptide we can facilitate synthesis by significantly shortening the amino acid sequence and by omitting the disulphide bridge. Most studies investigating defensin-based peptides have been focused on hBD-3, which is one of the most potent AMPs. By generating different amino-carboxy-terminal peptides, Hoover et al. identified several carboxy-terminal peptides with 9 to 14 amino acids having activity against E. coli or P. aeruginosa, but not against S. aureus 39 . In these peptides, cysteine residues were replaced by serine residues, which we found to be also crucial for activity in our peptides. Also, Reynold et al. described that antimicrobial activity of hBD-3 was mainly localized in the amino-terminal half 40 . Similar to our results, they reported that distinct amino acids are important for activity against different strains, suggesting that the strain-selectivity of such peptides can be modulated by varying the sequence.
To be utilized as potential antimicrobial molecules, it has to be excluded that the generated peptides provoke resistance of the treated microorganism. By using a host-derived antimicrobial peptide as therapeutic, this is of major importance, as resistance or cross-resistance towards other AMPs might be fatal for the host. While further testings are required to determine whether our identified peptides provoke such resistance, we believe that the octapeptide can serve as a backbone structure that could be optimized to enhance and/or specialize its activity and to diverge from its natural structure, which would decrease the risk of resistance induction. For example, due to their difference in activity against E. coli and C. albicans, optimization of the peptides RGKAKACK and RGKAKCAK (Fig. 1E) could generate a peptide that is effective against E. coli, but not against C. albicans.
In conclusion, we identified that the host can broaden its antimicrobial arsenal by generating several antibiotic molecules from the AMP hBD-1, depending on its redox state and proteolytic degradation. We believe that this strategy can be therapeutically exploited and that our identified hBD-1 derived carboxy-terminal peptides can be optimized for topical application against bacterial or fungal infections.

Materials and Methods
Bacterial and fungal strains. Bacterial  Radial diffusion assay. Antimicrobial radial diffusion assay was modified from reference 17 and performed as described earlier 11 . Briefly, microorganisms were cultivated (anaerobic bacteria with AnaeroGen, Oxoid, UK) for up to 18 hours in liquid TSB medium. Log-phase cultures were washed and diluted to 4 × 10 6 colony forming units in 10 ml agar. Incubation was carried out in 10 ml of 10 mM sodium phosphate, either pH 7.4 or 5.7, containing 0.3 mg/ml of TSB powder and 1% (w/v) low EEO-agarose (AppliChem) with 0 or 1 mM dithiothreitol (DTT, Sigma-Aldrich) under anaerobic or aerobic conditions for three hours. 1 or 4 µg of synthetic, oxidized hBD-1 (Peptide Institute, Japan) and 1 or 4 µg of synthetic peptides (EMC Microcollections, Tuebingen) were filled into small punched wells in a final volume of 4 µl. This concentrated peptide solution dilutes while diffusing into the gel, thereby generating concentration-dependent, round-shaped inhibition zones when killing immobilized microorganisms.
An overlay-gel containing 6% (w/v) TSB powder, 1% agarose and 10 mM sodium phosphate buffer without DTT was poured onto the plates and after incubation for up to 48 h at 37 °C the diameter of inhibition zones was measured. Experiments were repeated at least three times; mean + SEM is shown.

Microdilution broth assay.
To differentiate between microbistatic and microbicidal activity we performed a broth microdilution assay. For that, E. coli ATCC25922 bacteria were incubated overnight at 37 °C, 150 rpm. C. albicans ATCC 10231 was grown at 30 °C overnight, 150 rpm in liquid TSB. Cells were collected by centrifugation (2500 rpm, 10 min, 4 °C), washed twice and resuspended in 10 mM sodium phosphate buffer containing 1% (w/v) TSB broth. Required C. albicans cell density was adjusted using a hemocytometer. For bacteria the optical density of OD 600nm = 0.1 was determined. Approximately 5 × 10 5 CFU/ml bacteria or fungi were mixed with indicated peptide concentrations (1.25-200 µg/ml) in a final volume of 100 µl in 10 mM sodium phosphate buffer containing 1% (w/v) TSB broth and incubated for 2 hours at 37 °C. After incubation 10 µl per well were plated on LB-/ or YPD-agar plates to determine the CFU/ml. After that 100 µl of 6% TSB (w/v) were added and absorbance was measured at 600 nm (Tecan, Switzerland) and monitored for 18 hours. Growth relative to the positive control in % was plotted against peptide concentration. Experiments were carried out at least three times; mean ± SEM is shown.

Metabolic activity Assay.
To assess the metobalic activity of Caco-2 cells we used the WST-1 Cell Proliferation Reagent (Roche, Germany). Briefly, 1 × 10 5 Cells /ml were seeded and incubated with 100 µg/ml or 200 µg/ml octapeptides for 24 h at 37 °C, 5% CO 2 . After incubation the supernatant was removed and cells were washed with PBS and incubated with 20 µl Cell Proliferation Reagent WST-1 for 1 h 37 °C, 5% CO 2 . Finally the absorbance was measured at 450 nm and 620 nm. Experiments were repeated three times, mean+/− SEM.
Hemolytic Activity of antimicrobial peptides. An hemolytic activity assay for testing antimicrobial peptides was performed as described earlier 42 . Briefly, 150 µl of melittin (5 µM) was added to the positive control wells and incubated overnight. On the next day 1 ml blood was added to 3 ml PBS, mixed gently and centrifuged for 8 min, 700 × g. The supernatant was discarded and cells were re-suspended in 4 ml PBS and centrifuged again. After removing the supernatant, cells were centrifuged for 8 min at 1000 × g. Supernatant was discarded. For (2019) 9:3640 | https://doi.org/10.1038/s41598-019-40216-2 www.nature.com/scientificreports www.nature.com/scientificreports/ each well we used 75 µl of 1% Red blood cell (RBC) suspension in PBS. RBC suspension was mixed with indicated peptide concentration (2.5-200 µg/ml) in a final volume of 150 µl and incubated for 1 h at 37 °C. Finally the plate was centrifuged at 1000 × g for 10 min and 60 µl of supernatant was quickly transferred into a new plate. The absorbance was measured by 405 nm and 540 nm. Hemolytic activity was plotted relative to the 0.2% Triton X-100. Experiments were repeated three times; mean +/− SEM is shown. Culture of a model human oral epithelium. 1 × 10 6 TR146 cells (derived from a squamous cell carcinoma of the buccal mucosa; SkinEthic, France) were seeded into polycarbonate plastic inserts (Millipore) in DMEM (Lonza) and cultured for 8 days to form a multilayered epithelium. Medium in the wells (basal) and in the inserts (apical) was changed daily. On day 5 (airlift) medium was aspired from the apical side and cells were fed from the basal side for the rest of culture time in 6-well-plates. No antibiotics were used for the entire time of culture and the experiments 43 . The octapeptides were diluted to 100 µg/ml in 0.01% acetic acid and 50 µl of these dilutions were applied to the apical side of the model epithelia and incubated for 24 h. The supernatant was used for cytotoxicity assays and Enzyme-linked immunosorbent assay. Infection of model human oral epithelium with C. albicans. Octapeptides were diluted to 50 µg/ml in PBS, or an equivalent volume of 0.01% acetic acid were applied to the apical side of the RHOEs and incubated for 1 h. C. albicans SC5314 was synchronized as described previously 26 . Yeast cells were washed three times in PBS and 1 × 10 5 CFU were used for infection of pre-treated RHEs. After 24 h RHEs were fixed with Karnovsky's fixative. Post-fixed samples (1% OsO 4 , 1 h) were rinsed with distilled water, block-stained with uranyl acetate (2% in distilled water), dehydrated in alcohol (stepwise 30-96%), immersed in propylene oxide and embedded in glycide ether (polymerized 48 h at 60 °C, Serva, Heidelberg). Semi-thin sections were stained with toluidine blue and examined with a Nikon Eclipse 80i light microscope (magnification 1:400). Afterwards, fungal cells were simply coloured in red. Epithelial damage was evaluated by two independent experts in a blinded manner on a scale between 0 and 5 with 0 = intact epithelia, 1 = damage only in topmost cell layer; 2 = damage in top third of epithelium; 3 = damage in top half of epithelium; 4 = damage in all cell layers except lowermost; 5 = damage in all epithelial layers. ethics statement. The study protocol was previously approved by the Ethical Committee of the University Hospital, Tuebingen, Germany. Patients and controls who were included in this study all gave their written and informed consent after the study purpose, samples procedure, and potential adjunctive risks were explained. All experiments were conducted in accordance with the relevant guidelines and regulations.