Edinburgh Research Explorer The human cathelicidin LL-37 preferentially promotes apoptosis of infected airway epithelium

Cationic host defense peptides are key, evolutionarily conserved components of the innate immune system. The human cathelicidin LL-37 is an important cationic host defense peptide up-regulated in infection and inﬂammation, speciﬁcally in the human lung, and was shown to enhance the pulmonary clearance of the opportunistic pathogen Pseudomonas aeruginosa in vivo by as yet undeﬁned mechanisms. In addition to its direct microbicidal potential, LL-37 can modulate inﬂammation and immune mechanisms in host de-fenseagainstinfection,includingthecapacitytomodulatecelldeath pathways. We demonstrate that at physiologically relevant concen-trationsofLL-37,thispeptidepreferentiallypromotedtheapoptosisofinfectedairwayepithelium,viaenhancedLL-37–inducedmito- chondrial membrane depolarization and release of cytochrome c , with activation of caspase-9 and caspase-3 and induction of apopto-sis,whichonlyoccurredinthepresenceofbothpeptideandbacteria, but not with either stimulus alone. This synergistic induction of apoptosis in infected cells was caspase-dependent, contrasting with the caspase-independent cell death induced by supraphysiologic levels of peptide alone. We demonstrate that the synergistic in- duction of apoptosis by LL-37 and Pseudomonas aeruginosa required speciﬁcbacteria–epithelialcellinteractionswithwhole,livebacteria, and bacterial invasion of the epithelial cell. We propose that the LL-37–mediated apoptosis of infected, compromised airway epithelial cells may represent a novel inﬂammomodulatory role for this peptide in innate host defense, promoting the clearance of re- spiratory pathogens.

Cationic host defense peptides (CHDPs; also known as antimicrobial peptides) are key, conserved components of innate host defenses. The broad-spectrum, direct microbicidal potential of CHDPs has made these peptides attractive therapeutic agents. However, many CHDPs were further demonstrated to exert multiple potential immunomodulatory functions, including the modulation of cell death, raising questions about the nature of their primary physiologic roles and the possibility of developing novel therapeutics with both microbicidal and immunomodulatory activities.
One of the CHDPs demonstrating the most significant immunomodulatory potential to date is LL-37. This cationic, amphipathic peptide is the predominant cleavage product of human cationic antimicrobial peptide (hCAP)-18, the sole human cathelicidin (reviewed by Zanetti) (1). LL-37 is stored at high concentrations in the specific granules of neutrophils, and is produced by epithelial cells and some leukocytes. It can be detected in airway surface liquid, plasma, sweat, and other body fluids, and is upregulated in response to infection and inflammation (1,2). In addition to direct microbicidal capability, the modulatory potential of LL-37 is considerable, with in vitro and in vivo studies suggesting a broad range of activities that could modify innate inflammatory processes and adaptive immune responses (2). The physiological significance of LL-37 to human disease is demonstrated by the increased susceptibility to infection of individuals with morbus Kostmann (in which defective neutrophils are cathelicidin-deficient) (3), and is also suggested by the association between hCAP-18 expression and susceptibility to skin infections in psoriasis and atopic dermatitis (4). In addition, studies using a mouse model deficient in cathelin-related antimicrobial peptide (mCRAMP), the murine ortholog of LL-37, demonstrated increased susceptibility to infections of the skin, gastrointestinal system, urinary tract, and cornea (5)(6)(7)(8). Despite this clear evidence of a critical role for cathelicidin expression in innate defense against infection, the relative roles of the microbicidal and immunomodulatory activities of this peptide remain unclear.
Gene therapy augmentation demonstrated that the expression of LL-37 in the murine lung can enhance the clearance of pulmonary Pseudomonas aeruginosa (9), an important opportunistic pulmonary pathogen of immunocompromised individuals and those with cystic fibrosis (10). However, the mechanisms underlying enhanced defense against infection in this model remain unclear, with the concentrations of LL-37 detected unlikely to be directly microbicidal under physiological conditions (9,11). Multiple mechanisms are likely involved in the host defense against lung infection with P. aeruginosa, ranging from simple mucociliary clearance and innate microbicidal components of airway surface liquid, to the activity of professional phagocytes. In addition, the apoptosis and subsequent removal of infected epithelial cells were described as innate defense mechanisms at diverse epithelial surfaces (12)(13)(14), required for the clearance of invasive P. aeruginosa from the murine lung (12). Such a mechanism may be an important component of host defenses, removing bacteria that have evaded other defenses and invaded epithelial cells.
However, the roles of the Bcl2-family proteins, which can regulate mitochondrial membrane potential, and of the key apoptosis-inducing caspase proteins in LL-37, which can induce apoptosis of airway epithelial cells, remain uncertain. Further, it is unclear whether LL-37-induced apoptosis might be primarily detrimental, with overexpression of LL-37 damaging normal epithelial integrity, or whether at much lower, more physiological concentrations, LL-37 expression could enhance innate defenses by promoting targeted apoptosis to facilitate the clearance of pathogens. To address these issues, we studied the ability of LL-37 to induce apoptosis in airway epithelial cells infected with the invasive lung pathogen P. aeruginosa.
We demonstrate that LL-37 can induce Bax-dependent mitochondrial membrane depolarization in airway epithelial cells in a dose-dependent manner, with the release of cytochrome c, and that this is synergistically enhanced by infection with P. aeruginosa. However, at physiologically relevant concentrations of LL-37, the activation of caspase-9 and caspase-3, and DNA fragmentation, only occurred in the presence of both peptide and bacteria, but not with either stimulus alone. This synergistic induction of apoptosis was caspase-dependent and partly Bax-dependent, and required specific bacteria-epithelial cell interaction with whole, live bacteria capable of epithelialcell invasion.

Primary and Transformed Epithelial-Cell Culture
The 16HBE14o 2 transformed human bronchial epithelial cells were a kind gift from Dieter Gruenert at the University of California, San Francisco. Cells were grown in standard submerged cultured and maintained in DMEM supplemented with 1% L-glutamine (vol/vol), 1% NEAA (vol/vol), and 10% FBS (vol/vol). Culture flasks were coated with a basement layer of collagen IV (5 mg/ml), fibronectin (10 mg/ml), and BSA (100 mg/ml) before cell culture at 378C, 5% CO 2 . Normal human bronchial epithelial cells from a single donor were purchased from Lonza. Cells were grown in standard submerged culture and maintained in bronchial epithelial growth media (Lonza), supplemented with bronchial epithelial cell SingleQuots growth factors and supplements (Lonza) as a serum substitute, in strict accordance with the manufacturer's instructions.

Bacterial Strains and Culture
In addition to P. aeruginosa PAO1, these studies used the following strains of P. aeruginosa: clinical isolate J1386 (a clonal isolate of J1385, originally from an individual with cystic fibrosis) (23), DmexAB-oprM mutant (a gift from Keith Poole) (24), PAO1exsATV mutant (a gift from Dara Frank) (25), P. aeruginosa pilA mutant (a gift from Eva Lorenz) (26), and the isogenic PAO1 control strains for these mutants. Studies involving genetically modified bacteria were performed according to Scientific Advisory Committee on Genetic Modification Health and Safety Executive Certificate GM207/07.2.
All P. aeruginosa strains were grown in Luria Bertani (LB) broth at 378C in an orbital shaker (250 rpm) overnight, to achieve a stationary-phase suspension. Before use, bacterial suspensions diluted 1:20 in fresh LB broth were incubated at 378C for 90 minutes to reach log phase. Bacterial suspensions were standardized via dilution to an optical density of 0.1 at 595 nm, using spectrophotometry (WPA UV 1101, Biotech Photometer; Biochrom Ltd., Cambridge, UK), centrifuged at 1,500 3 g for 15 minutes (keeping supernatant where required for use in place of live bacteria), and resuspended in PBS before immediate addition to epithelial cells. Where required, bacteria were heat-killed (608C for 60 minutes in an orbital shaker) or ultraviolet light (UV)-killed (exposed to a constant UV source for 2 hours in a sealed glass Petri dish), with killing confirmed by overnight culture. To determine the direct microbicidal activity of LL-37, P. aeruginosa were resuspended in Ultroser G serum-substitute supplemented media before the immediate addition of LL-37 at the concentrations stated. After incubation for 1 hour at 378C, serial dilutions were performed, and 100-ml aliquots of these (and the original bacterial suspension) were spread onto LB agar plates in triplicate, and incubated overnight at 378C before counting the number of colony-forming units (CFUs). For studies to determine if the function of the DmexAB-oprM mutant could be rescued by soluble factors released by PA01, PA01 was added to 16HBE14o 2 cells at a multiplicity of infection (MOI) of 10:1 and incubated for 18 hours at 378C with 5% CO 2 . After incubation, the supernatant was collected and filtered through a 0.22-mm filter unit. The sterility of filtered supernatant was confirmed by culturing 50 ml on LB agar plates for 24 hours. Filtered supernatant (1:4 dilution in treatment medium) was simultaneously added together with LL-37 and DmexAB-oprM to 16HBE14o 2 cells, and the cells were incubated for 1 hour and analyzed using the mitochondrial depolarization technique.

Mitochondrial Depolarization Assay
The 16HBE14o 2 cells were seeded at 2.5 3 10 4 cells per well in a 96well plate and cultured at 378C, 5% CO 2 . Cells were exposed to LL-37 (or scrambled LL-37 control peptide) at the concentrations described in the presence and absence of (1) log-phase P. aeruginosa at an MOI of 10:1; (2) heat-killed or UV-killed bacteria (MOI 10:1), P. aeruginosa PAO1 LPS (1 mg/ml) or P. aeruginosa supernatant, all prepared as described above; or (3) log-phase P. aeruginosa PAO1 (MOI 10:1), separated from the epithelial cells by a Transwell semipermeable polyester membrane with 0.4-mm pore size (Corning Life Sciences, Amsterdam, Netherlands), and incubated for 1 hour at 378C, 5% CO 2 . All treatments were conducted in Ultroser G serum-substitute supplemented media. After treatments, cells were washed once with PBS prewarmed to 378C, the supernatant was aspirated, and 50 ml of Mitocapture solution (Cambridge Bioscience, Cambridge, UK) at 378C were added to each well, before incubation at 378C for 30 minutes in the dark. Cells were then washed twice with PBS at 378C, resuspended in 50 ml of Mitocapture buffer at 378C, and imaged immediately using an Axiovert S100 inverted fluorescent microscope (Zeiss UK, Welwyn Garden City, UK). For each membrane, at least four random fields of view were counted with a minimum of 300 cells in total, and the number of apoptotic cells (displaying diffuse, green fluorescence) was expressed as a percentage of the number of healthy cells (displaying punctate red mitochondrial fluorescence). Data were corrected for a background level of approximately 10% positive cells observed in control untreated samples. For inhibition studies, the culture medium in each well was replaced with treatment medium containing 100 mM BIP-V5 for 1 hour before treatment.

Cytochrome c Assay
The 16HBE14o 2 cells were seeded at 1 3 10 6 cells per well in a six-well plate and cultured at 378C, 5% CO 2 . Cells were exposed to LL-37 at the concentrations described in the presence and absence of log-phase P. aeruginosa PA01 (MOI 10:1), and incubated for 90 minutes at 378C, 5% CO 2 . Cells were washed once with PBS, and 0.5 ml of trypsin/ EDTA was added to each well to detach cells. Ultroser G serumsubstitute supplemented media (0.5 ml) was added to each well and suspensions were centrifuged in microtubes at 850 3 g for 2 minutes. Mitochondrial and cytosolic fractions were then prepared using a Mitochondrial Isolation Kit (Thermo Scientific, Loughborough, UK) according to the manufacturer's instructions. Cytochrome c concentrations in each fraction were assessed using a Cytochrome c ELISA Kit (Merck Chemicals, Ltd.), according to the manufacturer's protocol.

Western Immunoblotting
The 16HBE14o 2 cells were seeded at 1 3 10 6 cells per well in six-well plates and cultured at 378C, 5% CO 2 . Cells were exposed to LL-37 at the concentrations described, in Ultroser-G serum-substitute supplemented media, in the presence and absence of log-phase P. aeruginosa PA01 (MOI 10:1) and incubated for 3 hours at 378C, 5% CO 2 . Cells were washed once with ice-cold PBS and lysed with 300 mL M-PER protein extraction reagent (Thermo Scientific) containing a cocktail of protease, phosphatase, and metalloprotease inhibitors. Protein concentrations were determined by bicinchoninic acid assay (Thermo Scientific). Equivalent total protein concentration lysates (15-40 mg) were resolved on either 10% or 12% precast Precise Protein polyacrylamide gels (Thermo Scientific), transferred to polyvinylidene fluoride membranes (Bio-Rad Laboratories, Ltd., Hemel Hempstead, UK), blocked for 1 hour with Tris-buffered saline and 0.1% Tween-20 (TBST) containing 5% skimmed milk powder (TBST/milk), and then incubated with antibodies specific for cleaved caspase-3 (1 in 5,000 dilution), cleaved caspase-9 (1 in 1,000 dilution), XIAP (1 in 1,000 dilution), or pan-actin (1 in 2,000 dilution) in TBST/milk overnight at 48C. Membranes were washed for 15 minutes in TBST and then incubated with a 1 in 5,000 dilution of HRP-conjugated goat anti-rabbit antibody (in TBST/milk) or a 1 in 5,000 dilution of HRP-conjugated goat antimouse antibody (in TBST/milk) for 1 hour at room temperature. Membranes were washed for 30 minutes and developed with chemiluminescence peroxidase substrate (Sigma-Aldrich) according to the manufacturer's instructions, and imaged on CL-Xposure film (Thermo Scientific). Equal loading of protein was confirmed by examining the expression of actin as a loading control.

In Situ Cell Death Detection by Terminal Deoxynucleotidyl Transferase-Mediated Deoxyuridine Triphosphate Nick-End Labeling Assay
Transwell polyester-permeable supports (pore size, 0.4 mm; diameter, 6.5 mm; Corning Life Sciences) were equilibrated for 45 minutes in culture media (DMEM supplemented with 10% FBS [vol/vol], 1% L-glutamine [vol/vol], and 1% NEAA [vol/vol]) before the addition of 100 ml of medium containing 2.5 3 10 5 16HBE14o 2 cells/ml into the apical compartment, with 600 ml culture medium in the basal compartment, and cultured at 378C, 5% CO 2 . For primary bronchial epithelial cell experiments, Transwell supports were equilibrated with NHBE culture media for 45 minutes before the addition of 100 ml of NHBE media containing 2.5 3 10 5 NHBE cells/ml into the apical compartment, with 600 mL of NHBE culture medium in the basal compartment, and cultured at 378C, 5% CO 2 . Before treatments, culture media in both the apical and basal compartments were replaced with Ultroser G serum-substitute supplemented media. Cells were exposed to LL-37 at the concentrations described in the presence and absence of logphase P. aeruginosa PA01 (MOI 10:1), and incubated for 6 hours at 378C, 5% CO 2 . Cells were fixed in 10% neutral-buffered formalin (3.7% formaldehyde) for 10 minutes, washed once in PBS, permeabilized in ice-cold 0.1% Triton X-100/0.1% sodium citrate for 3 minutes, and washed twice with PBS. An in situ cell death detection kit (Roche Applied Science, West Sussex, UK) was used according to the manufacturer's instructions. The membranes with cells were mounted in 50 mL Vectashield Hardset (containing DAPI), and at least four random fields of view were counted (each containing more than 100 cells), using an Axiovert S100 fluorescent microscope, and analyzed using OpenLAB 3.0 software (Improvision/Perkin Elmer, Waltham, MA). The number of terminal deoxynucleotide transferase dUTP nickend labeling (TUNEL)-positive cells was expressed as a percentage of the number of DAPI-positive nuclei. The total number of DAPIpositive nuclei counted for each condition was determined, to evaluate total cell number. For inhibition studies, cells were prepared as described, and culture medium in each well was replaced with treatment medium containing either 100 mM Bax inhibiting peptide V5 (Merck Chemicals, Ltd.) or 50 mM of the broad-spectrum caspase inhibitor, Z-VAD-FMK (Merck Chemicals, Ltd.), for 1 hour before treatment.

Gentamicin Exclusion Assay
The capacity of P. aeruginosa isolates and mutants to invade epithelial cells was assessed using a gentamicin exclusion assay. Briefly, 16HBE14o 2 cells were exposed to strains of log-phase P. aeruginosa (MOI 10:1) for 60 minutes in Ultroser G-serum-substitute supplemented media. The media were removed from all wells, and cells were incubated with fresh media for 60 minutes with or without gentamicin (50 mg/ml) to kill extracellular bacteria. The media were then aspirated from gentamicin-treated cells, and these cells were vigorously washed with PBS and lysed with PBS containing 0.1% Triton X-100, and then plated on LB agar to determine internalized bacterial numbers. Media and/or epithelial cell lysates from wells without gentamicin were also plated on LB agar, to determine the number of associated bacteria and total infectious load. The CFUs were quantified by culturing overnight on LB agar plates at 378C.

Statistical Analysis
Statistical analyses were performed using Graphpad Prism version 5 for Windows (GraphPad Software Inc., La Jolla, CA). Statistical significance was assessed either using one-way ANOVA with Tukey's post hoc test, or two-way ANOVA with Bonferroni's post hoc test where appropriate. P < 0.05 was considered significant.

LL-37 and P. aeruginosa Synergistically Induce Epithelial Cell Death
To determine the capacity of LL-37 to induce cell death in infected airway epithelial cells, the human bronchial epithelial cell line 16HBE14o 2 was infected with P. aeruginosa PAO1, with or without concurrent exposure to LL-37. These cells were examined for nuclear DNA fragmentation by TUNEL assay at 6 hours ( Figure 1A). Pseudomonas aeruginosa alone did not induce cell death, and LL-37 alone induced cell death only at higher concentrations. However, concurrent exposure to both stimuli synergistically induced significant levels of cell death at greater than or equal to 20 mg/ml of LL-37 (P < 0.01), and even at an LL-37 concentration of 20 mg/ml that had no effect alone. A control scrambled LL-37 peptide had no effect. Total cell counts demonstrated no loss of cells during analysis (data not shown). To confirm these observation in nontransformed cells, primary human airway epithelial cells were used, and demonstrated the same response, with significant cell death induced in the presence of physiologically inflammatory levels of LL-37 only when infected with P. aeruginosa PAO1 ( Figure 1B).
To determine whether the cell death observed was apoptosis, cleavage of the key executioner caspase, caspase-3, was determined by Western immunoblot in 16HBE14o 2 cells at 3 to 6 hours after infection with P. aeruginosa PAO1, with or without concurrent exposure to LL-37. No activation was detected in response to LL-37 alone, or P. aeruginosa alone. However, concurrent exposure to both stimuli resulted in caspase-3 activation at 4 hours and thereafter ( Figure 1C and data not shown). These data indicate that the cell death induced synergistically by LL-37 and P. aeruginosa, but not by high concentrations of LL-37 alone, is caspase-dependent apoptosis. This finding is supported by the observation that preincubation with the polycaspase inhibitor Z-VAD-FMK significantly (P < 0.001) inhibited the synergistic induction of cell death by P. aeruginosa and LL-37 ( Figure 1A), reducing it to approximately the level induced by LL-37 alone at that concentration.
In addition, cleavage of caspase-9 (a key cytochrome c-activated initiator caspase) was also observed in response to infection with P. aeruginosa PAO1, only in the presence of LL-37 ( Figure 1D). Caspase-9 activation was not detected in response to LL-37 alone, or P. aeruginosa alone. In contrast, the activation of caspase-8 (a key death receptor-activated initiator caspase) was not evident (data not shown). These data demonstrate a synergistic induction of intrinsic apoptosisinducing pathways.
Given the absence of caspase-3 activation in response to concentrations of LL-37 at which peptide alone induced cell death, the expression levels of XIAP (a potent caspase inhibitor) were examined, but no effect on expression levels was evident ( Figure 1C).

Pseudomonas aeruginosa Infection of Airway Epithelial Cells Synergistically Enhances LL-37-Mediated Mitochondrial Depolarization and Cytochrome c Release
To determine the role of mitochondria in LL-37-induced cell death, 16HBE14o 2 cells were infected with P. aeruginosa PAO1, with or without concurrent exposure to LL-37. After 1 hour, the mitochondrial membrane potential was determined as an early indicator of apoptosis (Figure 2A). The LL-37 alone induced a dose-dependent increase in mitochondrial depolarization at greater than or equal to 20 mg/ml. Pseudomonas aeruginosa alone had no effect, but synergized with LL-37 to induce significantly greater mitochondrial depolarization than LL-37 alone, even inducing increased depolarization at low (10 mg/ml) LL-37 concentrations that had no effect alone (P < 0.05). Scrambled LL-37 peptide had no effect (Figure 2A). To determine whether this synergistic effect required an initial interaction between LL-37 and either the epithelial cell or the bacteria, which could subsequently alter bacteria-epithelial cell interactions, 16HBE14o 2 cells were infected with P. aeruginosa PAO1 for 1 hour, and washed before incubation with LL-37 for 1 hour. Under these conditions, the synergistic induction of mitochondrial depolarization was still evident, and even amplified at lower LL-37 concentrations ( Figure 2B). This result indicates that infection with P. aeruginosa promotes airway epithelial cell susceptibility to LL-37-induced apoptosis.
To evaluate the consequences of mitochondrial depolarization, the intracellular localization of cytochrome c was examined 90 minutes after 16HBE14o 2 cells were infected with P. aeruginosa PAO1, with or without concurrent exposure to LL-37 ( Figure 2C). The LL-37 alone induced a dose-dependent relocalization of cytochrome c from the mitochondria to the cytoplasm, reflecting the mitochondrial depolarization and TUNEL positivity observed, and reaching significance at 50 mg/ml LL-37 (P < 0.01). Pseudomonas aeruginosa alone had no effect, but synergized with LL-37 to induce a highly significant translocation of cytochrome c at all concentrations of LL-37 tested (P < 0.001). This latter effect was surprisingly pronounced, with very significant translocation observed even at 10 mg/ml of LL-37, a concentration at which significant cell death was not evident. Effects as yet unexplained on the mitochondria under these conditions (but not in response to peptide alone or bacteria alone) may have led to further translocation of cytochrome c from the mitochondria during sample preparation, with a resultant amplification of the effect observed. Cytoplasmic cytochrome c was detected by Western immunoblot in response to 10-30 mg/ml LL-37 only in infected cells (data not shown). Thus, the cytoplasmic translocation of cytochrome c was clearly evident under these conditions.
To determine whether the LL-37-mediated induction of apoptosis was dependent on the key proapoptotic Bcl-2 family protein Bax, the effects of exposure to LL-37 and P. aeruginosa on mitochondrial depolarization ( Figure 3A) and DNA fragmentation ( Figure 3B) were evaluated after preincubation with the Bax-inhibiting peptide V5 (BIP-V5). At high concentrations of LL-37, at which LL-37 alone induced substantial mitochondrial depolarization and apoptosis, the inhibition of Bax significantly (P < 0.01) and almost completely blocked these effects. In contrast, Bax inhibition only partly inhibited the combined effect of LL-37 and P. aeruginosa. These data demonstrate that caspase-independent induction of cell death by LL-37 alone is Bax-dependent. However, additional, and as yet unidentified, components are required for the synergistic enhancement of mitochondrial depolarization and induction of caspase-dependent apoptosis by LL-37 in P. aeruginosa-infected cells.

Synergistic Induction of Apoptosis by LL-37 and P. aeruginosa Requires Specific Bacteria-Epithelial Cell Interactions with Live Bacteria
To exclude the possibility that LL-37 exerted directly microbicidal effects on P. aeruginosa PAO1, bacterial viability was determined after exposure to LL-37 over the range of concentrations and in the culture media used for these studies ( Figure  4A). No significant, direct microbicidal activity was evident.
To examine whether the synergistic induction of apoptosis by LL-37 and P. aeruginosa required infection with live bacteria, and/or could result from secreted products, 16HBE14o 2 cells were exposed to a range of bacterial stimuli in the presence or absence of concurrent exposure to 30 mg/ml LL-37, and assessed for mitochondrial depolarization ( Figure 4B). The highly significant (P < 0.001), synergistic induction of mitochondrial depolarization observed after exposure to live P. aeruginosa and LL-37 was completely lost if the bacteria used were dead (heat-killed or UV-killed), or substituted with bacterial culture supernatant, or LPS prepared from PAO1 (1 mg/ml). Furthermore, physical separation of the epithelial cells from the bacteria by a semipermeable membrane also completely prevented this effect. These data indicate that the synergistic induction of apoptosis by LL-37 and P. aeruginosa requires a physical interaction between the epithelial cells and viable bacteria, and is not simply the result of pathogen-sensing by extracellular pattern recognition receptors.

Synergistic Induction of Apoptosis by LL-37 and P. aeruginosa Is Isolate-Specific, and Independent of Type III Secretion System and Pilus Expression
To exclude the possibility that the synergistic induction of apoptosis by LL-37 and P. aeruginosa was specific to PAO1, a clinical P. aeruginosa isolate J1386 was examined ( Figure 5A). A synergistic induction of mitochondrial depolarization was also observed in response to this clinical isolate in the presence of LL-37. This finding was substantially enhanced in comparison to that observed using the laboratory strain PAO1, with significant effects observed in infected cells after incubation with concentrations of LL-37 greater than or equal to 1 mg/ml (P < 0.01). No direct microbicidal effect of LL-37 was observed on P. aeruginosa isolate J1386 (data not shown).
To examine whether common virulence factors differentially expressed by divergent P. aeruginosa isolates were necessary for this effect, mutant strains of P. aeruginosa PAO1 were used ( Figures 5B and 5C). No substantial difference was evident when comparing an ExsA mutant with a defective type III secretion system (PAO1exsATV; Figure 5B) or a pilus mutant (pilA mutant; Figure 5C) with their corresponding isogenic strains. No direct microbicidal effect of LL-37 was evident in either mutant strain (data not shown). These data demonstrate that common determinants of virulence associated with epithelial-cell interactions (pilus) and bacterially induced epithelial cell death (type III secretion system) are not required for the synergistic induction of apoptosis in LL-37-treated infected epithelial cells.

Synergistic Induction of Apoptosis by LL-37 and P. aeruginosa Requires Epithelial-Cell Internalization of Bacteria
The internalization of P. aeruginosa by airway epithelial cells was proposed as a key component of the innate pulmonary host defense that is defective in cystic fibrosis (27). To determine the significance of bacterial internalization, a MexAB-OprM de-letion mutant (DmexAB-oprM), described as containing a defect in its ability to invade epithelial cells (28), was used. The 16HBE14o 2 cells were infected with PAO1DmexAB-oprM, or the isogenic control, and a gentamicin-exclusion assay was performed to determine the extent of internalization into the epithelial cells ( Figure 6A). Whereas internalization of the isogenic strain could be clearly demonstrated, negligible internalization of the PAO1DmexAB-oprM bacteria occurred. Furthermore, the LL-37-mediated synergistic enhancement of apoptosis was significantly (P < 0.001) and completely lost when using PAO1DmexAB-oprM ( Figure 6B). The invasion defect of PAO1DmexAB-oprM was previously shown to result from the absence of a bacterial secreted factor, and can be restored by the addition of supernatants from isogenic control bacteria exposed to epithelial cells (28). The LL-37-mediated synergistic enhancement of apoptosis was significantly restored (P < 0.01) when 16HBE14o 2 cells were infected with PAO1DmexAB-oprM in the presence of both LL-37 and supernatants from isogenic control bacteria exposed to epithe-lial cells ( Figure 6C). These data demonstrate that the epithelial-cell internalization of P. aeruginosa is required to facilitate the LL-37-mediated induction of apoptosis at physiologically inflammatory concentrations of peptide.

DISCUSSION
Cationic host-defense peptides, including LL-37, have been demonstrated to have multiple properties capable of modulating inflammation and immunity. The full extent of these properties remains to be determined, but understanding the physiological roles of CHDPs in health and disease, and their development as antimicrobial therapeutics, is clearly significant. Our results suggest a novel innate inflammomodulatory role for LL-37, preferentially inducing apoptosis in infected epithelial cells, with the potential to exert protective or detrimental effects.
The most critical mechanisms by which cathelicidins contribute to host defense against infections remain uncertain. In various models, LL-37 and mCRAMP exert antimicrobial effects in vivo (5-9), despite high minimum inhibitory concentration values that often exceed detectable physiologic levels. Recent studies implicated the vitamin D-dependent up-regulation of LL-37 in the intracellular killing of mycobacteria in mononuclear leukocytes (29,30), perhaps in synergy with b-defensin 4 (31), and mCRAMP impairs the intracellular replication of Salmonella (32). Therefore, these peptides likely have direct antimicrobial roles where peptides are concentrated in favorable, controlled ionic conditions, and perhaps function synergistically with other agents. However, the function of LL-37 at epithelial surfaces, at the peptide concentrations reported, is less clear. In lungs, hCAP-18 was detected in bronchoalveolar lavage fluid from healthy infants at approximately 5 mg/ml, and was found at up to approximately 25 mg/ml and at approximately 15 mg/ml in infants with pulmonary infections and individuals with cystic fibrosis lung Figure 2. Pseudomonas aeruginosa infection of airway epithelial cells synergistically enhances LL-37-mediated mitochondrial depolarization and cytochrome c release. Human bronchial epithelial cells (16HBE14o 2 ) were incubated with a range of LL-37 concentrations (or scrambled LL-37  at 50 mg/ml) in Ultroser G serumsubstitute supplemented media, in the presence and absence of logphase P. aeruginosa PA01 (MOI 10:1). Bacteria and LL-37 were added concurrently and incubated for 60 minutes (A) or 90 minutes (C), or epithelial cells were preinfected with bacteria for 60 minutes, washed, and exposed to LL-37 for 60 minutes (B). (A, B) Mitochondrial membrane depolarization was determined using Mitocapture dye, quantifying the percentage of apoptotic cells displaying diffuse green fluorescence (cells with depolarized mitochondria), compared with healthy cells displaying punctuate red fluorescence (cells with polarized mitochondrial membranes). Four random fields of view were counted for each sample (minimum of 300 cells per sample), and number of apoptotic cells was expressed as a percentage of the total number of cells. Data were corrected for a background level of approximately 10% positive cells in control untreated samples, and plotted as mean values 6 SEM, for n 5 6 (A) or n 5 3 (B) independent experiments for each condition. Two-way ANOVA with Bonferroni post hoc test was performed to determine significance. *P < 0.05, **P < 0.01, ***P < 0.001. (C ) Cellular localization of cytochrome c was assessed by ELISA analysis of mitochondrial fractions after subcellular fractionation. Data represent the mean percentage of cytochrome c present in this fraction as a proportion of total cytochrome c detected in each sample 6 SEM for n 5 3 independent experiments, measured in duplicate for each condition. Two-way ANOVA was performed with Bonferroni post hoc test to compare each treatment to appropriate LL-37-free negative control sample. **P < 0.01, ***P < 0.001. b disease (in steady state), respectively (33,34). At these sites, immunomodulatory roles may be of primary significance. Indeed, the potential significance of such activities was recently demonstrated in terms of the in vivo protection against infection in animal models using a synthetic CHDP derivative with no direct antimicrobial activity in vitro (35).
A variety of CHDPs, including bovine cathelicidins and human a-defensins, were shown to affect eukaryotic cell death (18,36). We previously showed that high (potentially supraphysiologic) concentrations of LL-37 induced apoptosis in pulmonary epithelial cells in vitro and in vivo (15,17). However, the mechanisms involved remain undetermined. We demonstrate here that at these higher concentrations, LL-37 can induce mitochondrial depolarization and cytochrome c release . LL-37-induced mitochondrial depolarization and DNA fragmentation involve Bax-dependent mechanisms. Human bronchial epithelial cells (16HBE14o 2 ) were incubated for 1 hour (A) or 6 hours (B) over a range of LL-37 concentrations in Ultroser G serum-substitute supplemented media, in the presence and absence of log-phase P. aeruginosa PA01 (MOI 10:1) added concurrently, with or without preincubation for 1 hour with Bax-inhibiting peptide V5 (BIP-V5; 100 mM). (A) Mitochondrial membrane depolarization was determined using Mitocapture dye, quantifying the percentage of apoptotic cells displaying diffuse green fluorescence (cells with depolarized mitochondria), compared with healthy cells displaying punctuate red fluorescence (cells with polarized mitochondrial membranes). Four random fields of view were counted for each sample (minimum of 300 cells per sample), and the number of apoptotic cells was expressed as a percentage of total number of cells. Data were corrected for a background level of approximately 10% positive cells in control untreated samples, and plotted as mean values 6 SEM, for n 5 3 independent experiments for each condition. A two-way ANOVA with Bonferroni post hoc test was used to compare LL-37-only treated samples with LL-37/BIP-V5treated samples, or LL-37/P. aeruginosa-treated samples with LL-37/P. aeruginosa/BIP-V5-treated samples at corresponding concentrations. *P < 0.05, **P < 0.01, ***P < 0.001. (B) Cells were fixed and apoptosis was assessed by TUNEL assay. Four random fields of view, each containing more than 100 cells, were counted for each sample, and the number of TUNEL-positive cells was expressed as a percentage of the number of DAPI-positive nuclei. Data represent mean values 6 SEM, for n 5 3 independent experiments for each condition. Twoway ANOVA with Bonferroni post hoc test was used to compare LL-37 only-treated samples with LL-37/BIP-V5-treated samples, or LL-37/P. aeruginosa-treated samples with LL-37/P. aeruginosa/BIP-V5-treated samples at corresponding concentrations **P < 0.01, ***P < 0.001. Figure 4. Synergistic induction of apoptosis by LL-37 and P. aeruginosa requires specific bacteria-epithelial cell interactions with whole, live bacteria. (A) P. aeruginosa PA01 was cultured to log-phase, then exposed to LL-37 over a range of concentrations for 1 hour at 378C in Ultroser G serum-substitute supplemented media. Serial dilutions were performed, incubated on LB agar plates in triplicate, and cultured for 16 hours before colony-forming units were counted. Data represent mean values 6 SEM, for n 5 3 independent experiments for each condition. (B) Human bronchial epithelial cells (16HBE14o 2 ) were assessed for mitochondrial membrane depolarization using Mitocapture dye, as described in MATERIALS AND METHODS, after incubation for 1 hour with a range of concentrations of LL-37, in serum-substitute supplemented media, in the presence and absence of live log-phase P. aeruginosa PA01 (MOI 10:1), heat-killed or UV-killed PA01 (MOI 10:1), P. aeruginosa PA01 LPS (1 mg/ml), P. aeruginosa PA01 conditioned medium, or live P. aeruginosa PA01 (MOI 10:1) separated from the cells via a semipermeable polyester membrane with 0.4-mm pore size. Data represent mean values 6 SEM, for n 5 3 independent experiments for each condition. Two-way ANOVAs were performed to evaluate significance, with Bonferroni post hoc tests comparing LL-37 alone to LL-37/stimuli. ***P < 0.001.
in airway epithelial cells, confirming previous findings in alveolar epithelial cells (18). In addition, the LL-37-mediated induction of mitochondrial depolarization and the subsequent apoptosis of these cells can be completely blocked using the BIP-V5 peptide inhibitor of the proapoptotic Bcl-2 family protein Bax. The BIP-V5 peptide mimics the Bax-binding domain of Ku70, preventing Bax translocation from cytosol to the mitochondria (37). This translocation is a central event in mitochondria-dependent apoptosis, with the subsequent activation and oligomerization of Bax and Bak resulting in either the nonspecific rupture of, or the formation of specific channels in, the outer mitochondrial membrane and release of cytochrome c (38). Interestingly, we demonstrate that the Bax-dependent LL-37-mediated release of cytochrome c did not cause an activation of caspase-3 or caspase-9 after exposure to LL-37 alone, yet resulted in a Bax-dependent DNA fragmentation. In addition, polycaspase inhibition resulted in only a partial inhibition of the apoptosis induced by high levels of LL-37 (15). These data suggest that the induction of apoptosis by high concentrations of LL-37 alone appears to be a Bax-dependent and predominantly caspase-independent process, and may implicate the liberation and activation of mitochondrial apoptosis-inducing factor (AIF) and/or endonuclease G. The mechanism by which LL-37 can interact with or activate Bax in airway epithelial cells is unclear. LL-37 could induce an opening of the mitochondrial permeability transition pore, as proposed for bovine myeloid antimicrobial peptide-28 (BMAP-28) (36). However, a study published during preparation of our manuscript described a calpain-dependent mechanism of LL-37-mediated Bax translocation to the mitochondria, responsible for the AIF-mediated apoptosis induced by very high concentrations (50-200 mg/ml) of LL-37 in Jurkat T leukemia cells. These findings are compatible with our data (21). Irrespective of this, we demonstrate that concentrations of LL-37 considered to be physiologically relevant during lung inflammation (10-30 mg/ml) induce minimal apoptosis in human airway epithelial cell lines and primary cells, in the absence of infection. This result suggests that under normal physiological conditions, LL-37 on epithelial surfaces would not be damaging.
In contrast to the effects of LL-37 alone, cells infected with P. aeruginosa demonstrated an enhanced susceptibility to the induction of apoptosis upon exposure to concentrations of LL-37 that had no effect alone, but not to control scrambled LL-37 peptide. This effect comprised a pronounced synergistic increase in mitochondrial depolarization, cytochrome c release, and DNA fragmentation, and was at least partly Bax-independent. In addition, the LL-37-mediated activation of caspase-3 and caspase-9 was evident only in infected cells, demonstrating activation of the intrinsic pathway of apoptosis. Although P. aeruginosa infection alone has been shown to induce extrinsic pathways of apoptosis via CD95/CD95L (12), we saw no activation of capase-8 and no significant cell death in response to bacteria alone in our system. This finding may relate to the fairly low MOI used, and the timeframe examined in our studies, suggesting that the LL-37-mediated induction of apoptosis in infected epithelial cells is a much earlier (and mechanistically distinct) form of cell death compared with previously described, bacterially induced death receptor-mediated apoptosis. The intrinsic pathway of apoptosis is a mitochondrialdependent mechanism of caspase activation involving cytochrome c-induced oligomerization of the cytosolic apoptotic protease activating factor-1 (Apaf-1), which recruits and activates procaspase-9, an upstream activator of effector caspases, such as caspase-3 (39). In addition, the mitochondrial release of Smac/DIABLO (second mitochondrial activator of capases/ direct IAP binding protein with low PI) (40) and Omi (also known as high temperature requirement factor A2 [HtrA2]) (41) leads to an inactivation of the inhibitor-of-apoptosis proteins (IAPs) that normally inhibit caspase activity. The increased apoptosis observed via TUNEL assay in infected cells exposed to LL-37 could be inhibited by the polycaspase in- Figure 5. Synergistic induction of apoptosis by LL-37 and P. aeruginosa is isolate-specific and independent of type III secretion system and pilus expression. Human bronchial epithelial cells (16HBE14o 2 ) were assessed for mitochondrial membrane depolarization using Mitocapture dye, as described in MATERIALS AND METHODS, after incubation for 1 hour with a range of concentrations of LL-37, in Ultroser G serumsubstitute supplemented media, in the presence and absence of (A) log-phase clinical P. aeruginosa isolate J1386 (MOI 10:1), (B) log-phase P. aeruginosa PA01exsATV or isogenic PAO1 control strain (MOI 10:1), and (C) log-phase pilA P. aeruginosa mutant or isogenic PAO1 control strain (MOI 10:1). Data represent mean values 6 SEM, for n 5 3 independent experiments for each condition. Two-way ANOVAs were performed to evaluate significance, with Bonferroni post hoc tests comparing (A) LL-37/P. aeruginosa to LL-37 alone, and (B) LL-37/P. aeruginosa mutant to LL-37/isogenic controls. *P < 0.05,***P < 0.001. hibitor Z-VAD-FMK, reducing it to levels similar to those induced by LL-37 alone. Thus the synergistic effects are caspasedependent, and occur in addition to predominantly caspaseindependent pathways induced by higher concentrations of LL-37 alone. The caspase inhibition by IAPs may be reduced in infected cells, and although XIAP levels were unaffected, the roles of Smac/DIABLO and Omi HtrA2 in this system remain unknown. Therefore, a caspase-dependent pathway downstream of mitochondrial depolarization, induced by an alternate mechanism from that used by LL-37 alone, is responsible for the capacity of LL-37 to promote the apoptosis of cells infected with P. aeruginosa.
The nature of the interaction between epithelial cells and bacteria required to make these cells susceptible to the apoptosis-inducing effects of LL-37 was investigated under a number of conditions and using mutants of P. aeruginosa. Neither dead bacteria nor soluble products produced by untreated or LL-37-treated bacteria could promote these synergistic effects. In the absence of physical contact between the epithelial cells and live bacteria, no effects were observed. In contrast, the effect of LL-37 was even more profound when a clinical strain of P. aeruginosa J1386 (isolated from an individual with cystic fibrosis) (23) was used, suggesting that this effect might be modified by isolate variation in virulence factors. PAO1 is classified as an ''invasive'' rather than ''cytotoxic'' strain of P. aeruginosa (although both can invade eukaryotic cells), and this invasiveness is proposed to require contact between bacteria and epithelial cells to stimulate the efflux of bacterial ''invasive factors'' (28). The DmexAB-oprM deletion mutant of P. aeruginosa PAO1 (24) is defective in terms of epithelialcell invasion (despite normal adherence), and has diminished virulence in vivo as a consequence of the loss of the MexAB-OprM efflux system, proposed to be responsible for the efflux of these putative ''invasive factors'' (28). A synergistic induction of apoptosis was not evident in LL-37-treated epithelial cells infected with this mutant strain, but could be replicated by the addition of these unknown ''invasive factors'' from the isogenic wild-type PAO1 strain, demonstrating a requirement for invasiveness. In contrast, the PAO1exsATV mutant (25), in which the ExsA mutation impairs the ExsA-regulated type III secretion system, behaved identically to its isogenic wild-type PAO1 strain. Although a functional ExsA allele is required for P. aeruginosa-induced cytotoxicity, epithelial-cell invasiveness is independent of ExsA expression (42). Similarly, a P. aeruginosa pilA mutant (26) was largely able to synergize with LL-37 to induce apoptosis as effectively as its isogenic PAO1 wild-type strain. In this strain, pilA mutation results in an absence of pilus, proposed to be an important adhesin involved early in epithelial-cell interactions with P. aeruginosa (43). Interestingly, differences were observed in the sensitivity to the LL-37induced mitochondrial depolarization of cells infected with our original PAO1 isolate, compared with isogenic controls for some of the mutants used. Additional investigations using these isolates may help in further defining the key events involved in this interaction. Nevertheless, the data suggest that the bacterial invasion of airway epithelial cells, but not ExsAregulated type III secretion or pili expression, is critical in inducing enhanced susceptibility to LL-37-mediated apoptosis.
Our results describe a novel innate inflammomodulatory role for LL-37, preferentially inducing the apoptosis of infected epithelial cells. However, the extent to which this might contribute to innate epithelial defenses, or be manifest in pathologic damage to epithelial-barrier integrity, is unknown, and a fine balance could exist. Although LL-37 clearly has important roles in innate host defense against infection, chronically increased hCAP-18/LL-37 concentrations in cystic fibrosis Figure 6. Synergistic induction of apoptosis by LL-37 and P. aeruginosa requires epithelial-cell internalization of bacteria. Human bronchial epithelial cells (16HBE14o 2 ) were incubated for 60 minutes in Ultroser G serum-substitute supplemented media, in the presence and absence of (MOI 10:1) log-phase P. aeruginosa strains PA01, DmexAB-oprM mutant (A-C ), isogenic PAO1 control strain (B), or DmexAB-oprM mutant added concurrently with sterile conditioned supernatant collected from 16HBE14o 2 cells infected with PA01 (C ). (A) Invasion of epithelial cells by bacteria was determined by gentamicin exclusion, quantifying the number of viable CFUs surviving extracellular gentamicin treatment (50 mg/ml). Data are plotted as mean values 6 SEM, for n 5 3 independent experiments plated in duplicate for each condition. (B, C ) Infected epithelial cells were concurrently incubated with a range of concentrations of LL-37, and mitochondrial membrane depolarization was determined using Mitocapture dye, as described in MATERIALS AND METHODS. Data represent mean values 6 SEM, for n 5 3 independent experiments for each condition. Two-way ANOVAs were performed to evaluate significance, with Bonferroni post hoc tests comparing (B) LL-37/DmexAB-oprM mutant to LL-37/isogenic controls, and (C ) LL-37/DmexAB-oprM mutant to LL-37/DmexAB-oprM mutant in PAO1-conditioned supernatant. **P < 0.01,***P < 0.001. lung disease are correlated with increased lung damage (34), and elevated hCAP-18/LL-37 concentrations are associated with bronchiolitis obliterans syndrome (44) and the pathogenesis of psoriasis (45). Pulmonary epithelial-cell apoptosis plays a significant role in P. aeruginosa clearance from the murine lung (12). In addition, bladder epithelial-cell exfoliation after bacterial attachment plays a role in innate defense against invasive Escherichia coli (14), preventing the establishment of a safe niche and intracellular biofilm-like growth (46). Furthermore, the susceptibility of individuals with cystic fibrosis to pulmonary P. aeruginosa infection is proposed to relate, in part, to the failure of airway epithelial cells to internalize this bacterium, and thus an inability to clear P. aeruginosa by desquamation of infected cells (27). Thus, we propose that in the healthy host, LL-37, up-regulated during infection and inflammation, may promote the apoptosis and consequent clearance of P. aeruginosa-infected airway epithelial cells, as a component of the innate host defense against this pathogen. However, under pathologic conditions of excessive, chronic LL-37 exposure, or a failure of epithelial-cell internalization of P. aeruginosa (such as in cystic fibrosis), the epithelial-cell death induced by high concentrations of LL-37 alone may be detrimental to the host and contribute to chronic lung damage. The extent to which this effect might be common to other invasive bacteria, or else specific to P. aeruginosa, remains to be determined, but has clear significance for the possible use of LL-37 and related CHDPs as antimicrobial therapeutics.