Isolation and characterization of Human ß-Defensin-3, a novel human inducible peptide antibiotic

and cloned from including hBD-3-treated hBD-3 (expressed as a His-Tag-fusion protein in E. as well as chemically synthesized hBD-3 were indistinguishable from naturally-occurring peptide with respect to their antimicrobial activity and biochemical properties. Investigation of different tissues revealed skin and tonsils to be major hBD-3 mRNA expressing tissues. Molecular cloning and biochemical analyses of antimicrobial peptides in cell culture supernatants revealed keratinocytes and airway epithelial cells as cellular sources of hBD-3. Tumor necrosis factor alpha and contact with bacteria were found to induce hBD-3 mRNA expression. hBD-3 therefore

Recent investigations revealed that α-defensins also have the ability to attract T cells (21).
Very recent investigations indicate that human ß-defensins attract immature dendritic cells and memory T cells via its chemokine receptor CCR6 (22) providing a link between innate epithelial defense and adaptive immunity.
Whereas skin infections caused by Gram-negative bacteria are rather rare, S. aureus is a major cause for skin-and lung infections, in particular in atopic dermatitis (23). The high abundance of hBD-2 in skin (16) might explain its high resistance against Gram-negative bacterial infection. In contrast, the factors that protect skin from S. aureus-infection remain speculative.
We therefore hypothesized that human skin produces peptide antibiotics directed against S. aureus in addition to the Gram-negative bacteria killing hBD-2.
In the present study, we report the discovery of a novel human epithelial broad spectrum and multi-resistent bacteria killing peptide antibiotic, which we termed human beta-defensin-3 (hBD- 3), that is inducibly expressed by various human epithelial cells.

Experimental Procedures
Culture of Epithelial Cells. Foreskin-derived keratinocytes, airway epithelial cells and the A549 lung epithelial cells were prepared and cultured as described previously (17,24). Purification and characterization of hBD-3. Pooled lesional psoriatic scales (7 -50 grams) or heel callus (80 -120 grams) were extracted with acidic ethanolic citrate buffer as described previously (25). After diafiltration (Amicon filters, cut off: 3 kDa) of extracts (or the supernatants of cultured epithelial cells) against sodium phosphate buffer (10 mM, pH 7.4), material was applied to a S. aureus affinity column, which was prepared using a NHS-activated sepharose column (HiTrap, 5 ml; Pharmacia) and 5 ml of a S. aureus (clinical isolate) suspension (10 9 bacteria/ml) by a procedure similar to that previously described for a Ps. aeruginosa affinity column (17). Briefly, extracts or cell culture supernatants were applied to the affinity column that had been previously equilibrated with 10 mM phosphate buffer, pH 7.4, and bound peptides were eluted with 0.1 M glycine buffer, pH 3.0, containing 1M NaCl. After equilibration of the column with 10 mM phosphate buffer, pH 7.4, the effluent was applied to the column and bound material was eluted as described above. This step was performed four times to increase the efficacy of the column to bind peptides. The eluates were collected and diafiltered against 0.1% trifluroacetic acid (TFA), pH 3, for subsequent RP-HPLC.
Fractions containing antimicrobial activity against S. aureus were further purified by cation exchange HPLC followed by RP-18-HPLC as described for purification of hBD-2 (17).
Electrophoretic mobility was investigated using SDS-polyacrylamide gels (SDS-PAGE) in the presence of 8M urea and Tricine (27) under nonreducing conditions as described for chemokines (28). Peptides were visualized by silver staining (27).
Protein sequencing was done using a pulsed liquid phase 776 automated protein sequencer (Perkin Elmer Applied Biosystems). Electrospray mass spectrometry (ESI-MS)-analyses were performed in the positive ionisation mode with a QTOF-II Hybrid-mass spectrometer (Micromass).

Antimicrobial/hemolytic assay.
Test organisms were incubated with hBD-3 in 100 µl 10 mM sodium phosphate buffer (pH 7.4) containing 1% (v/v) trypticase soy broth (TSB). To investigate the salt sensitivity of hBD-3, 50 µg hBD-3 was incubated with 1 x 10 5 colonyforming units (CFU) of S. aureus (ATCC 6538) in 100 µl 10 mM sodium phosphate buffer (pH 7.4) and NaCl for 3 hours at 37 0 C . The antibiotic activity of hBD-3 was analyzed by plating serial dilutions of the incubation mixture and determination of the CFU the following day. The limit of detection (1 colony per plate) was equal to 1x10 2 CFU per milliliter.
For analysis of hemolytic activity up to 500 µg hBD-3 were incubated with 1x10 9 /ml human erythrocytes at 37 0 C for 3 hrs either in 10 mM sodium phosphate buffer (pH 7.4) containing 0.34 M sucrose or only in phosphate buffered saline. Following incubation samples were centrifuged at 10000 x g for 10 min and hemolysis was determined by measuring the A 450 of the supernatants using 0.1% Triton X-100 for 100% hemolysis. 5% phosphate-buffered glutaraldehyde (pH 7.8) for 2 h, repeatedly rinsed in cold phosphate buffer, and postfixed for a further 2h in 4% phosphate-buffered osmic acid. The sample was dehydrated in acetone and finally embedded in Araldit (Araldit Cy212, Sigma), as described previously (29). Bacteria were examined with an EM 910 electron microscope (Zeiss). For determination of hBD-3 mRNA in different tissues total RNA was isolated from human skin, larynx, pharynx, polyp, tonsil and tongue using the TRIzol reagent (Gibco-BRL). All other RNAs were obtained from Clontech (Palo Alto, CA). Real-time RT-PCR was carried out as described above.

Expression of recombinant hBD-3 in E coli
The cDNA encoding the 45 amino acids containing natural form of hBD-3 was cloned into the expression vector pET-30c (Novagen), which contains an N-terminal His-Tag sequence allowing purification of the fusion-protein by the use of a nickel-affinity column.

Results
Isolation of a novel human peptide antibiotic: hBD-3. To address the question whether human skin produces S. aureus killing proteins, we analyzed lesional scale extracts of patients with psoriasis and extracts of healthy person's heel callus for S. aureus killing activity. Initial experiments revealed high S. aureus killing activity in crude psoriatic scale extracts as well as heel stratum corneum extracts. In order to enrich and purify staphylocidal activity from psoriatic scale extracts, in which more staphylocidal activity was observed than in heel callus extracts, a S.
aureus-affinity column was used. Protein(s) with microbicidal activity directed against S. aureus were found to bind to the column. Bound proteins were then separated by preparative reversedphase-C 8 high performance liquid chromatography (HPLC) and HPLC-fractions were analyzed for staphylocidal activity (Fig. 1A). The most prominent staphylocidal activity containing HPLC-fraction was further purified using micro cation exchange HPLC.
Staphylocidal activity eluted at high salt concentration from this column indicating a highly basic antimicrobial peptide (Fig. 1B, arrow). Final purification of this antibiotic peptide was achieved by reversed phase C 2 /C 18 HPLC (Fig.1C). Tricine-SDS-urea-polyacrylamide gel electrophoretic analyses revealed a single line migrating like a 9 kDa polypeptide (Fig. 1C, inset). NH 2 -terminal amino acid sequence analyses gave the sequence shown in Fig. 1D (EMBL/Genbank database, accession number P 81534), which indicated a new human antimicrobial peptide. Using degenerated primers the complementary DNA (cDNA) was isolated from primary keratinocytes. The cDNA (EMBL/Genbank database, accession number AJ237673) encodes a 67 amino acid precursor and the predicted 45 amino acids containing mature peptide shows similarity to vertebrate epithelial ß-defensins, in particular bovine "enteric ß-defensin", EBD (Fig. 2). Since this novel antimicrobial peptide is the third isolated human ß-defensin it was termed human ß-defensin-3 (hBD-3).
By electrospray mass spectrometry its exact molecular mass was found to be 5154.59 Da, which is 6 Da less than the mass calculated from the deduced hBD-3 amino acid sequence (5161.20 Da), supporting the idea that hBD-3 contains three cysteine bridges and the amino acid sequence shown in Fig. 2.
We were able to isolate 88 µg pure hBD-3 from 7 grams psoriatic scales and 15 µg hBD-3 from 112 grams human skin-derived stratum corneum. The recovery of hBD-3 peptide after three HPLC purification steps was found to be very low. Losses were estimated to be in the range of 80 -95 % of the quantity originally present.
hBD-3 peptide could be expressed as His-Tag-fusion protein in E. coli. Cleavage of the fusion protein, which was found to be weakly active against E. coli, with Enterokinase and subsequent RP-HPLC-analysis led to a single peptide giving by ESI-MS a molecular mass of 5154.2 Da, which is exactly the mass calculated for full length hBD-3. Antimicrobial activity against E. coli (Fig. 3B) and Staphylococcus aureus was found to be equivalent to that seen for natural hBD-3. Synthetic hBD-3 gave a single peak by RP-HPLC at the same retention time as natural hBD-3 and gave a molecular mass of 5154.7 Da upon ESI-MS analyses.

hBD-3 exhibits salt-insensitive broad
S. aureus was killed by hBD-3 at low and physiologic salt concentrations (Fig. 3C). Reduced antimicrobial activity was only observed at supraphysiological salt concentrations.
Since several cationic antimicrobial peptides have been reported to exhibit cytotoxic activity against eukaryotic cells, hBD-3 was also assayed for hemolytic activity against human erythrocytes. No significant hemolytic activity (< 0.5 %) was observed using concentrations of hBD-3 up to 500 µg/ml at physiologic salt concentrations. However, significant hemolytic activity was seen at high hBD-3 concentrations in 10 mM sodium phosphate buffer containing 0.34 M sucrose (Fig. 3A).

Ultrastructure of hBD-3-killed Staphylococcus aureus.
In order to develop an insight into the mechanisms by which S. aureus is possibly killed by hBD-3, we examined the morphological changes of S. aureus exposed to hBD-3 by transmission electron microscopy. As shown in Fig respectively, induced hBD-3 mRNA (Fig. 5B-C).

hBD-3 peptide is produced by keratinocytes and lung epithelial cells.
We then investigated, whether epithelial cells produce hBD-3 peptide. Biochemical analyses of culture supernatants of primary keratinocytes as well as A549 lung epithelial cells previously pretreated with Ps. aeruginosa led to the isolation of a peptide antibiotic showing identical biochemical properties including the N-terminal sequence as seen for the skinderived hBD-3 (data not shown). We were able to purify approximately 10 µg hBD-3 from the supernatants of both, 10 9 primary keratinocytes and 10 9 A549 cells indicating that skin keratinocytes as well as epithelial cells of the respiratory tract represent cellular sources for hBD-3.
Although in human secretions such as tears secretory phospholipase A2 may represent one of the most potent Gram-positive bacteria killing factors (39), no systematic analyses have been performed to elucidate why human healthy skin is protected from S. aureus infection. Our previous observation, that hBD-2 is not bactericidal towards S. aureus (15,17) prompted us to investigate human skin extracts for S. aureus killing factor(s). These analyses have led to the purification of a novel peptide antibiotic, which we identified as human beta-defensin 3 (hBD-3). In order to elucidate how S. aureus is possibly killed by hBD-3, we examined morphological changes occurring upon hBD-3-treatment of S. aureus by transmission electron microscopy.
The morphological effects resemble those seen when S. aureus is treated with penicillin, an antibiotic that interferes with the crosslinking of the bacterial peptidoglycan cell wall (42).
Therefore, mechanisms by which hBD-3 affects S. aureus seem to be completely different from those discussed for neutrophil α-defensins where lamellar mesosome-like structures at the cell membrane level were seen in S. aureus (43). Striking electron-dense deposits were present in the periplasmic space affixed to the outer membrane when E. coli was investigated (44). It has been suggested that antimicrobial peptides bind and insert into the cytoplasmic membrane due to their cationic and amphiphilic characteristics, where they assemble into multimeric pores (45). However, a very recent investigation indicates, that -at least in the case of the octamer-forming hBD-2 -bactericidal activity could also result from electrostatic charge-based mechanisms of membrane permeabilization, rather than a mechanism based on formation of bilayer-spanning pores (46). It remains to be determined whether hBD-3 kills bacteria by a similar mechanism and how hBD-3 affects cell wall perforation in S. aureus.
The identification of hBD-3 in normal stratum corneum and the isolation of hBD-3-peptide from culture supernatants revealed skin keratinocytes as a possible cellular source of hBD-3.
Expression of hBD-3 in epithelial cells was further confirmed by the detection of hBD-3 mRNA in primary keratinocytes as well as in primary respiratory epithelial cells, where we also isolated the protein from culture supernatants. Whereas low hBD-3 mRNA expression was found in many normal epithelial tissues including that of the respiratory tract and genitourinary tract, real-time RT-PCR revealed high levels of hBD-3 mRNA-expression in skin and surprisingly, tonsils. It is interesting to speculate, that possibly microbial stimulation is responsible for these findings.
The isolation of 10-to 30-fold higher amounts of hBD-3 from psoriatic lesions, when compared with normal stratum corneum, indicated that hBD-3 is also inducible by inflammatory stimuli. Like hBD-2 (16,18) and the epithelial bovine ß-defensins LAP and TAP (47) and unlike hBD-1 (48), proinflammatory cytokines such as TNF-α induce hBD-3 in primary epithelial cells at physiologically relevant concentrations. Furthermore the contact of epithelial cells with bacteria induces hBD-3 gene expression, a finding that is known for hBD-2 in keratinocytes (15), airway epithelial cells (17) and intestinal epithelium (38). Thus hBD-3 represents the second member of the human beta-defensin family where expression is regulated by inflammatory stimuli at a transcriptional level.
Several reports indicate that inactivation of antimicrobial peptide activity in patients with cystic fibrosis (CF) may contribute to the recurrent airway infections (49). Elevated salt concentrations in the airway surface fluid of patients with CF, a matter that has been controversially discussed (49), inactivate the antimicrobial activity of human ß-defensins (13,18,35), possibly by inhibiting the binding of positively charged defensins to negatively charged bacterial surfaces. In contrast to both known human ß-defensins (18,35), our findings indicate that bactericidal activity of hBD-3 is not salt-sensitive at physiologic salt concentrations , which makes this ß-defensin of particular relevance in CF.
In summary, the discovery of a novel human epithelial broadspectrum antimicrobial peptide confirms the hypothesis that antimicrobial peptides represent an integral part in the innate immunity of human epithelia (as is found in organisms lacking an adaptive immune system (50)) which complements the adaptive cellular immune system and offers an immediate host response against infectious agents. After we had submitted the protein sequence (P81534) and the cDNA sequence (AJ237673) for hBD-3, sequencing of a chromosome 8 BAC clone (AF189745) revealed the presence of a nucleotide sequence encoding a putative ß-defensin-like protein identical to hBD-3. Thereafter two cDNA sequences were available in the GenBank database, which encode hBD-3 (Accession number AF217245 and AF295370).