Published ahead of print on December 30, 2002, doi:10.1165/rcmb.2002-0191OC
© 2003 American Thoracic Society DOI: 10.1165/rcmb.2002-0191OC The Antimicrobial Activity of the Cathelicidin LL37 Is Inhibited by F-actin Bundles and Restored by GelsolinDivision of Pulmonary Medicine, The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania; and Department of Physiology, University of Pennsylvania, Institute for Medicine and Engineering, Philadelphia, Pennsylvania Address correspondence to: Paul A. Janmey, Dept. of Physiology, University of Pennsylvania, Institute for Medicine and Engineering, 1010 Vagelos Research Laboratories, 3340 Smith Walk, Philadelphia, PA 19104. E-mail: janmey{at}mail.med.upenn.edu
Antimicrobial peptides are part of the innate host defense system, and inactivation of these peptides is implicated in airway infections in cystic fibrosis (CF). The sputum of patients with CF contains anionic polyelectrolytes, including F-actin and DNA not found in normal airway fluid. These anionic filaments aggregate to contribute to the altered viscoelastic properties of CF sputum. We hypothesized that the airway components stabilizing bundles of F-actin and DNA are in part cationic antimicrobial agents, and that appropriate modification of diseased airway fluid of patients with CF might dissociate these bundles and restore antimicrobial activity. We demonstrate that the human cathelicidin peptide LL37 forms bundles with F-actin and DNA, which are dissolved by gelsolin and DNase, respectively. Coincident with bundle formation, antimicrobial activity of LL37 is inhibited by F-actin and DNA. Pseudomonas bacteria were killed by low concentrations of LL37, but killing was significantly reduced in the presence of F-actin. The actin filament–fragmenting protein gelsolin restored bactericidal activity nearly completely. In a growth inhibition assay, the effects of F-actin were confirmed, and DNA was also shown to inhibit the activity of LL37. When added to CF sputum, gelsolin significantly reduced the growth of bacteria, suggesting activation of endogenous antimicrobial factors. These findings may have therapeutic implications for treatments previously thought to alter only the viscoelastic properties of airway secretions and amplify the possible advantage of gelsolin in CF treatment.
Abbreviations: cystic fibrosis, CF • CF transmembrane conductance regulator, CFTR • dynamic light scattering, DLS • interleukin, IL • room temperature, RT
Defects in the cystic fibrosis (CF) transmembrane conductance regulator (CFTR) lead to altered secretions from exocrine glands (including those in the airways), increased sputum viscosity, a heightened susceptibility to airway infections, and severe airway inflammation (1). Most patients with CF eventually succumb to respiratory failure as a consequence of an unrelenting cycle of inflammation and infection. Over 80% of patients with CF are chronically infected by Pseudomonas aeruginosa, frequently with isolates resistant to antibiotics (2). In addition, the normal host defense of the lung may be defective in CF (3), despite apparently normal levels of endogenous antimicrobial agents. As survival improvement in the last several years has plateaued, new tools are still needed to treat CF lung disease. More recent tools include new strategies to treat antibiotic-resistant pathogens (4) as well as mucolytic agents able to dissolve CF sputum. In addition to altered ionic and hydration states, the airway fluid in CF also contains significant amounts of the anionic polymers F-actin (5) and DNA (6) that are released from neutrophils and other cells lysed as the result of inflammation. Polyelectrolytes with the high surface charge density of DNA and F-actin have strong interactions with oppositely charged polyvalent cations, even in solutions of physiologic or elevated monovalent salt concentrations. When sufficient concentrations of polyelectrolytes and multivalent counterions are present, the polyelectrolytes collapse into bundles that trap both polyelectrolytes and counterions in dense, stable structures (7). Because all known airway antimicrobial peptides and most clinically used antibiotics are polyvalent cations, these agents are likely to interact with F-actin and DNA in CF sputum, and as has also been documented for the polycationic interleukin (IL)-8 (8), such interactions can block the biological activity of these compounds. To test the hypothesis that anionic polyelectrolytes present in CF airway fluid might inhibit the function of natural and pharmacologic antibiotics, we tested the interactions of these agents with F-actin and DNA in vitro, and examined the antibacterial function of some of these antimicrobial compounds in systems containing F-actin or DNA with and without agents that depolymerize these filamentous polymers. Our data demonstrate that F-actin bundle formation can be induced by a range of structurally distinct antibacterial agents including LL37, ß-defensins, lactoferrin, and tobramycin. Addition of F-actin to LL37 leads to loss of antibacterial function. Moreover, actin bundling by LL37 is reversed by gelsolin, an actin-fragmenting protein that has been considered for clinical use in CF and other lung diseases. Dissolution of F-actin/LL37 bundles restored bacterial killing activity to these purified systems. We also show that DNA bundle formation can be induced with LL37, and overcome with DNase. The hypothesis that polyelectrolyte concepts may guide therapeutic strategies is supported by evidence that gelsolin has the expected ability to increase bacterial killing in actin-containing sputum samples derived from patients with CF.
Antimicrobials LL37 and mouse ß-defensin 3 were synthesized using an automatic peptide synthesizer monitored by mass spectrometry and capillary electrophoresis at the Louisiana State University Medical Center Core Laboratories. Tobramycin sulfate (Eli-Lilly, Indianapolis, IN) and colymycin (Colistimethate; Monarch Pharmaceuticals, Bristol, TN) were obtained from our hospital pharmacy and diluted from the stock concentration of the intravenous preparations. Lactoferrin from human milk was obtained from Sigma (St. Louis, MO).
F-Actin Preparation
Measurement of F-Actin and DNA Bundle Formation
Microbroth Dilution Assay
Bactericidal Assay
Expectorated Sputum
Bundle Formation The formation of F-actin and DNA bundles was detected by changes in light scattering. Figure 1A shows the bundle formation of 2 µM F-actin induced by LL37 peptide at concentrations 0–10 µM. The minimal bundling concentration of this antimicrobial agent is similar to the concentration required for effective bacteria killing (Figure 2A). The kinetics of F-actin bundling after mixing 2 µM F-actin with 10 µM LL37 peptide are shown in Figure 1B. The time course of bundling by LL37 is similar to that induced by other polycations (10) and suggests a progressive increase in both bundle size and concentration. When gelsolin is added to F-actin, to reduce the length of the filaments, no increase in scattering is observed after adding LL37. When gelsolin is added after bundles have been formed by addition of LL37 to long actin filaments, the scattering intensity is abruptly decreased to the level of unbundled F-actin. These results suggest that the increased scattering is not due to direct interactions between LL37 and actin monomers producing insoluble aggregates, and is consistent with electrostatically-driven filament bundling, where a critical filament length is required for bundling to occur (10). The concentration dependence of gelsolin's ability to prevent LL37-driven actin bundling is shown in Figure 1C. A significant inhibition of bundling is observed at a gelsolin:actin ratio of 1:20, and inhibition is complete above a 1:10 ratio.
LL37 was also effective as a DNA bundling agent in both isotonic and hypotonic conditions (Figures 1D and 1E). Similar to the effect of gelsolin on F-actin bundles, DNase treatment dissolved DNA/LL37 peptide bundles, but DNase digestion required a longer time (Figure 1E) consistent with the sequential digestion of DNA by enzymatic amounts of DNase compared with the rapid stoichiometric reaction of gelsolin with F-actin. To test if LL37 interferes with DNase activity, fragmentation of DNA was measured by gel electrophoresis. Figure 1F shows that digestion of DNA by DNase I is not prevented by coincubation with LL37 under conditions where DNA aggregation would occur. To explore whether the actin/DNA bundling property of LL37 might be a general feature of antimicrobial peptides or clinically used antibiotics, we conducted similar bundling experiments with lactoferrin (Figure 3A), mouse ß-defensin 3 (Figure 3C), and two clinically used antipseudomonal antibiotics: tobramycin, an aminoglycoside antibiotic with maximally five positive charges, and colymycin, a polypeptide antibiotic with no net charge (Figures 4A and 4B). The F-actin crosslinking activity of lysozyme, another antibacterial agent of the airway fluid, has previously been documented (11). Figures 3A and 3C show that lactoferrin and ß-defensin 3 also cause the association of F-actin into assemblies that scatter more light. Consistent with the hypothesis that the filament aggregation is driven by electrostatic effects, bundling is reversed or inhibited by addition of soluble multivalent anions, such as polyaspartate, that compete with the F-actin for the cationic ligands and prevent the self-association of actin filaments (Figures 3B and 3C). As shown in Figure 4A, tobramycin also induces F-actin bundling in hypotonic medium at concentrations above 1 µM, and tobramycin-induced DNA bundle formation occurs in both hypotonic and isotonic media with thresholds greater than 1 µM and 10 µM, respectively. No F-actin or DNA bundle formation was observed with colymycin at the same concentration range (Figure 4B).
Effects of F-Actin and DNA on Antimicrobial Activities of Peptides The activity of LL37 was initially explored using a standard colony-forming unit assay in which bacteria are incubated with antimicrobials for 1 h, and then plated onto standard culture media agar plates; live bacteria are then quantified after an overnight incubation. In Figure 2B, it can be seen that LL37 killed nearly all (95%) PAO1 organisms at relatively low concentrations (1.4 µM). In the presence of F-actin, only 21% of the bacteria were killed by 1.4 µM or 2.75 µM LL37, and concentrations of 11 µM were required to kill 94% of the bacteria. Gelsolin added to the LL37/actin mixture increased the number of bacteria killed to 53% at 1.4 µM and 96% at 2.75 µM LL37. The effect of F-actin was confirmed using the microbroth dilution assay, which measures growth inhibition rather than acute killing activities of antimicrobials. In this assay, LL37 was again shown to inhibit growth at concentrations of 1.4–2.75 µM, resulting in only small increases in optical density of the culture (representing bacterial growth) when compared with media alone, over the 18-h incubation (Figure 2A). F-actin (2 mg/ml) nearly completely suppressed the antimicrobial effect of LL37 at low medium concentrations (1.4–6 µM); no antimicrobial activity was seen until the concentration of LL37 was raised to 10 µM. DNA (1 mg/ml) ablated antimicrobial activity at all tested concentrations (Figure 2A). To determine if the effects of polyanions were specific to LL37, experiments were performed using antibiotics clinically used in the treatment of pulmonary infections in patients with CF. Consistent with the finding that the minimal bundling concentration of tobramycin and colymycin (Figures 4A and 4B) are higher than the concentration ensuring its effective bacteria-killing property, we do not observe a loss of tobramycin and colymycin antibacterial activity against PAO1 in the presence of F-actin (4 mg/ml) or DNA (1 mg/ml) using the growth inhibition assay (Figures 5A and 5D).
Cystic Fibrosis Sputum Expectorated sputum samples from three patients with CF were incubated with gelsolin, DNase, or LL37, singly or in combination, to determine whether the bacterial load of the sputum was altered by these agents. Western blot analysis of three sputum samples (Figure 6, inset) demonstrated intact actin as well as smaller degradation products, consistent with previous findings (14). Both gelsolin and DNase decreased the bacterial load by  50% when compared with untreated sputum (Figure 6). Because neither gelsolin nor DNase are bactericidal in assays with bacteria alone (Figure 2), their effects in sputum suggest liberation of endogenous antibacterial function within the sputum. When gelsolin was added together with exogenous LL37 or DNase, there was a slight apparent decrease in bacterial growth, the statistical significance of which will require additional studies.
The principle finding of this study is that DNA and actin undergo bundle formation in the presence of cationic antimicrobials, and this bundling impairs the bacterial killing activity of these peptides. The antimicrobial activity of cationic peptides can be recovered from actin or DNA bundles using gelsolin or DNase. We also demonstrate that DNase or gelsolin treatment can alter the bacterial burden of sputum from patients with CF. The mechanisms by which mutations in the CFTR protein cause persistent airway infection in patients with CF remain poorly understood. One leading hypotheses (12) suggests that absence of CFTR-mediated interaction with the epithelial sodium channel results in hyperabsorption of sodium with subsequent passive movement of water, resulting in depletion of water at the apical surface. This effect results in severe alterations in mucociliary clearance, and presumably allows for small numbers of bacteria to colonize and infect the lower airways. In addition to altered salt and water balance, it has long been recognized that sputum from patients with CF contains large amounts of DNA (3–14 mg/ml) derived from lysed inflammatory cells (8, 13). Filamentous actin is also released from inflammatory cells to yield concentrations of 0.1–5 mg/ml in CF sputum, and likely contributes to the properties of the thickened sputum (5, 14). F-actin and DNA are among the most highly charged polyelectrolytes in the cell, and normally they are complexed with proteins either bearing a net positive charge such as histones, or having a polycationic actin– or DNA-binding site. The equilibrium of such electrostatically-mediated binding events is shifted when cytoplasmic contents enter the extracellular fluid, where multiple cationic species including antimicrobial agents, each present at micromolar or greater concentrations, displace the less abundant or less cationic intracellular ligands for F-actin and DNA.
Aerosolized recombinant human DNase (rhDNase) has been used for a decade in the United States as a treatment aimed at improving airway clearance of thickened secretions in patients with CF (15, 16). In a phase III clinical trial in 968 patients with CF, it had a modest positive effect on lung function (5.8% improvement in FEV1) and risk of pulmonary exacerbations (28% reduction in age-adjusted risk) after 24 wk of treatment. The medication is costly ( In addition to its effects on airway secretion viscoelastic properties, anionic DNA has been shown to bind cationic proteins such as neutrophil elastase (8, 19). CF sputa treated with gelsolin or DNase demonstrated increased IL-8 bioavailability (8), suggesting that activity of this cationic amphiphile is sequestered by the polymeric anions. Release of the neutrophil chemoattractant IL-8 from actin or DNA may be detrimental to the inflammatory milieu of the CF airway. Furthermore, patients with non-CF idiopathic bronchiectasis had more pulmonary exacerbations and a greater decline in lung function when treated with rhDNase (20). The release of IL-8 (and possibly other inflammatory cationic proteins) from DNA bundles may offer one explanation of why DNase treatment does not improve lung function in all patients. Reassuringly, Costello and coworkers (21) found that adult patients with CF treated with rhDNase for 12 wk had lower sputum and plasma elastase activity. Shah and colleagues (19) found that IL-8 and neutrophil elastase were both increased in the sputum of patients treated with rhDNase, but these increases were transient, and over a 6-mo follow-up period, both markers decreased. In addition to DNA, CF sputa have been found to have large amounts of filamentous actin that increases the viscosity and the elastic modulus by more than 100% (5). Gelsolin, a protein able to sever actin filaments, was shown to decrease the viscosity of CF sputum in vitro (5). In addition to the direct contribution of actin to sputum viscosity, actin can bind to DNase, presumably decreasing the efficacy of this drug in the airway. Gelsolin can dissociate actin from DNase (22), resulting in enhanced activity of the enzyme to remove free DNA from the airway. In this way, gelsolin and DNase may act synergistically to decrease viscosity of airway secretions. The filamentous DNA and F-actin released into the airway fluid are mostly found in the form of large bundled aggregates containing both types of anionic polymer (14). The formation of such polyelectrolyte bundles can be mediated by a variety of polyvalent cations normally found in airway fluid or released there as the result of inflammatory cell function or necrosis. Among these cationic proteins and peptides are the antibiotic factors of the lung, including ß-defensins, cathelicidins, lactoferrin-derived peptides, and lysozyme. The experiments reported in this study show that one consequence of this bundle formation is loss of antibacterial function, but that in the case of F-actin, depolymerization of the filament by gelsolin dissolves the bundles and liberates the antibacterial function lodged within. This finding suggests that strategies to disassemble the DNA/F-actin aggregates in CF sputum may restore the antimicrobial function otherwise inhibited by the charged polymers. In addition to depolymerizing the filaments by gelsolin or DNase, the electrostatically-driven bundles can also be dissociated by soluble multivalent anions such as polyaspartate (Figure 3B). The formation of polyelectrolyte complexes depends very strongly on small changes in valence and polyion concentration, though it is only weakly affected by monovalent salt. Preliminary experiments to restore antibacterial activity by dissolving LL37/F-actin bundles with polyaspartate have not yet been successful, perhaps because the addition of polyaspartate itself can act as a multivalent counterion for LL37 that inhibits bactericidal activity. Further work to optimize conditions may lead to restoration of biological activity. The much weaker or insignificant interaction of antibiotic drugs with F-actin and DNA suggests that these drugs may function in contexts where the natural antimicrobials are inhibited. The stoichiometry of the charge interaction appears to be important, and high doses of drug appear to partially overcome the inhibition by bundling. That is, high concentrations of cationic antimicrobial peptides might be able to overcome the inhibitory effects of DNA or actin bundles. Indeed, the aim of aerosolized delivery of drugs such as antibiotics is to achieve very high local (airway) concentrations. In the case of tobramycin, specifically, it is estimated that aerosolized delivery can result in local concentrations in the airway lining fluid as high as 200 µg/ml. In our studies, tobramycin induced F-actin bundling only in low tonicity, but was able to promote bundling of DNA in low or high tonicity. Thus, whatever the answer to the controversial question about the actual concentration of salt in the CF airway surface liquid, this bundling phenomenon may occur. Colymycin, an uncharged peptide antibiotic, did not promote bundling of F-actin or DNA (Figure 4). In summary, our data demonstrate that DNA or actin forms bundles in the presence of cationic antimicrobials, and thereby impairs the killing activity of these peptides. We can only speculate whether this effect occurs in vivo, and whether some of the beneficial effects of DNase treatment observed in patients are due to enhanced antimicrobial activity in addition to the effects on the viscoelastic properties of CF sputum. Our in vitro experiment using expectorated CF sputum suggests that rhDNase can alter the bacterial burden of the sputum, although there have been no studies to date to address whether rhDNase affects the microbiology of the CF airway fluid in vivo. Our data suggests that this mechanism of action is deserving of study. We conclude that DNase and gelsolin may both have therapeutic effects on the function of cationic antimicrobial peptides, mediated by their effects on DNA or F-actin bundle formation.
D.J.W. is supported by the Harry Schwachman Clinical Investigator award from the Cystic Fibrosis Foundation. R.B. is supported by post-doctoral fellowship based on CFF grant R881. P.A.J. is supported by grants from the Cystic Fibrosis Foundation and NIH HL67286. The authors gratefully acknowledge the help of James. M. Wilson and Christian Moser in facilitating this study, and the assistance of Marianne Ferrin, MSN, and patients of the Adult Cystic Fibrosis Center of the University of Pennsylvania, for providing sputum samples.
* These authors contributed equally to the work presented in this article.  Received in original form September 19, 2002 Received in final form December 18, 2002
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
Abstract
Introduction
0.3) in 5 ml of tryptic soy broth (Becton-Dickinson, Cockeysville, MD) in a shaking incubator (300 rpm, 37°C). One milliliter of the bacterial suspension was centrifuged at 4,000 rpm for 5 min at RT, and the bacterial pellet was resuspended in 0.25x Muller Hinton Broth (MHB; Becton-Dickinson). This suspension was diluted 1:1,000 in 0.25x MHB, and used for the assay. Serial dilutions of antibiotic (LL37, tobramycin, or colymycin) were prepared in 0.01% (vol/vol) acetic acid containing 0.1% (vol/vol) bovine serum albumin. The antimicrobials were mixed with diluted bacterial suspension and F-actin. In some experiments, the mixture was also treated with gelsolin. In each experiment, positive controls (bacteria with no additives) and negative controls (media with no bacteria) were included. The total volume of each mixture was 100 µl, and was incubated in a 96-well plate at 37°C for 18 h, after which the OD600 was measured. 
