INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Interleukin-8 (IL-8) is a member of the -chemokine superfamily and the principal neutrophil chemoattractant in the CF airways (1). It is synthesized as a peptide of molecular weight 8.3 kD, yet it exists in the airways predominantly in the form of macromolecular complexes with molecular weights as great and greater than 1,000 kD (2). In vitro experiments have shown that heparin, a strongly anionic polymer, binds to the small cationic peptide IL-8 by what are believed to be electrostatic interactions, and we have previously shown an inhibition of neutrophil chemotactic activity by this mechanism (3). Analysis of the composition of CF sputum reveals large quantities (3 to 14 mg/ml) of the viscous anionic polymer DNA (4, 5) and actin (0.1 to 5 mg/ml) (6), which are released by necrosing neutrophils (7) after their recruitment into the airways during the IL-8-mediated inflammatory response.

In the neutrophil, as in all living eukaryotic cells, DNA exists in complex with binding proteins, the histones and the nonhistone chromosomal proteins, of which the histones form the vast majority, with their mass being roughly equivalent to that of the DNA itself. The histones are highly positively charged, and it is this feature, as opposed to nucleotide sequence, that governs their binding to the negatively charged DNA double helix. These charge interactions help to stabilize the DNA and to pack it efficiently within the cell nucleus since, without the histones, the length of DNA strands would span the nucleus several thousand times. The overall charge effect of the cationic histones on the DNA molecule is termed "cationic shielding" (8).

Necrosis is an unordered process, and the behaviour of DNA after necrosis of the neutrophil in the CF airways may determine several properties of airway secretions. During necrosis, the nuclear and cell membranes degrade and chromatin is spilt randomly into the surrounding tissue, unlike the ordered events of apoptosis in which chromatin is degraded by an endonuclease within the cell (9). Once the chromatin is released in the tissue, the histone component is subject to degradation by enzymes, including neutrophil elastase (10), the protease that is found at unusually high concentrations and whose activities appear central to much of the CF airway pathology (11). The DNA itself is also subject to a degree of degradation by endogenous DNase (7). As the histones are released, their condensing properties are lost, so that the DNA expands to fill its uncoiled volume. Secondly, in the absence of its cationic shield, the anionic surface of the DNA molecule is exposed. The relatively large volume of polyanionic material that can be supposed to arise as a result of these activities may not only contribute to increased viscosity of the CF airway secretions but may also attract cations such as the basic chemokines present in the airways. Both of these features have been ascribed to extracellular DNA in the CF airways. Similarly, F-actin is a polyelectrolyte that forms bundles in the presence of basic polypeptides (12). A previous report suggested that IL-8 does not bind to DNA (13), although this was not substantiated with data, and the concept that IL-8 may bind to actin has not previously been investigated. We now hypothesize that these polyanions do bind IL-8 and that this is relevant to the therapeutic use of DNA and actin depolymerizing agents in CF.

Concentrations of DNA and polymeric F-actin are related to sputum viscosity, and their depolymerization, by DNase I or gelsolin, respectively, is generally associated with a decrease in viscosity and implied improvement in mucociliary clearance and pulmonary function (6, 14, 15). Inhaled human recombinant DNase I (Pulmozyme) is currently prescribed to more than 40% of patients with CF in the United States and a smaller but steadily increasing proportion across Europe (Roche Pharmaceuticals, Welwyn Garden City, UK, communication from company and [16]). It was also suggested that gelsolin be investigated as a potential mucolytic agent in patients with CF (7). Reports that DNase I in vivo has no effect on sputum IL-8 concentration (13, 17, 18) have failed to distinguish between IL-8 that is bound to macromolecules and that which is free and therefore biologically active. Extracellular DNA in the airways is known to bind the basic proteases, cathepsin G and neutrophil elastase (19), as well as certain cationic antibiotics that appear to be released after DNase treatment (20). We have therefore conducted in vitro experiments to demonstrate IL-8 binding to DNA and actin and release of free IL-8 from CF sputa by the mucolytics DNase I and gelsolin.

The hypothesis that IL-8 binds to either DNA or actin was tested by investigating the affinity of IL-8 for heparin in the presence and absence of these polyanions. In addition, IL-8 incubated with either DNA or actin was separated on the basis of molecular weight by gel filtration chromatography. The interactions between IL-8, DNA, and DNase were investigated by treating CF sputa in vitro with DNase. Bovine DNaseI was used to test this principle as it shares 77% homology with human DNaseI and acts by the same mechanism, relying on divalent cations for optimal enzymatic activity. Similarly, the interactions between IL-8, actin, and gelsolin were investigated by treating the sputa with gelsolin. IL-8 was measured in its free and complexed forms, before and after DNase or gelsolin treatment, by two specific ELISA, and functional analysis of the sputa was determined in PMN chemotaxis assays. The contribution of IL-8 bioactivity in the sputa was determined in the presence of an IL-8 neutralizing antibody. Finally, the effects of DNA or actin on the binding of IL-8 to neutrophils was measured using radiolabeled IL-8 in the presence and absence of either polyanion.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

DNA and Actin IL-8 Binding Assays

Human recombinant IL-8 (50 µg/ml; gift of Dr I. Lindley, Novartis, Vienna) was incubated overnight at 37° C, either with phosphate-buffered saline (PBS: GIBCO, Paisley, UK) alone, or mixed with human placental DNA, 5 mg/ml (Sigma Chemical Co., Poole, Dorset, UK) in PBS, or porcine muscle G-actin 0.5 mg/ml (Sigma), in distilled water, in a final volume of 1 ml. Subsequent analysis by gel filtration chromatography (see below) demonstrated that, under these conditions, the G-actin became polymerized. Each incubated mixture was fractionated either by heparin affinity or by gel filtration chromatography using the Bio-Rad Biologic system: 1 ml of the mixture was applied to a heparin affinity column (Econocolumn; Bio-Rad Laboratories, Herts, UK) equilibrated with 0.05 M TRIS HCl (pH, 7.4), and eluted with a NaCl concentration gradient (0 to 2 M) in this TRIS buffer. Sequential 0.5 ml fractions were collected and assayed by immunoblot (see below). Then 250 µl of each mixture were applied separately to a gel filtration column (Superdex 200; Pharmacia, Uppsala, Sweden) and eluted at 0.5 ml/min in Hanks' buffered saline solution (HBSS; GIBCO). Sequential 0.25-ml fractions were collected and assayed by immunoblot for IL-8.

Immunoblot Analysis

Fractions (3 µl) were applied to 0.4 µm nitrocellulose membrane (Bio-Rad) and air-dried. The membranes were blocked with PBS/1% Tween-20 for 1 h at room temperature. They were then incubated overnight at room temperature with a goat antihuman IL-8 (5 µg/ml; gift of Dr. I. Lindley, Novartis, Vienna), or antihuman actin (1:1,000; Sigma) in PBS/0.1% Tween-20. After two 5-min washes in PBS/0.1% Tween-20, the sheets were incubated with biotinylated rabbit antigoat IgG (1:1,000; Dako, High Wycombe, UK) in PBS/0.1% Tween-20 for 90 min at room temperature. Sheets were again washed twice in PBS/ 0.1% Tween-20, and streptavidin-biotin-peroxidase complexes were added at 1:1,000 for 90 min at room temperature. After two final washes as above, the immunoblots were stained with diaminobenzidine (0.5 mg/ml; Sigma) in PBS/0.1% Tween-20 containing 0.05% hydrogen peroxide.

Sputum Treatment

CF sputum samples were treated with either DNase I, gelsolin, or buffer controls. Ten CF sputum specimens were collected by spontaneous expectoration from patients attending routine pediatric outpatient clinics (age, 13 [8 to 18] yr, FEV₁ > 40% predicted, eight female patients) at Southampton General Hospital, and stored at 20° C prior to treatment. Sputum DNA measurements (7.33 ± 2.3 mg/ml) were made using Hoechst 33258 (21) and were within previous estimates (4, 5). Each sample was split into four equal aliquots and treated with bovine DNaseI (100 µg/ml; Sigma) in 0.15 M NaCl, gelsolin (250 nM; Sigma) in 0.15 M KCl, 20 mM TRIS at pH 7.6, 0.2 mM CaCl₂, 0.2 mM ATP, 1 mM DTT, or buffer controls. The samples were incubated for 4 h at 20° C and centrifuged at 20,000 × g for 20 min. The supernatants were aspirated and stored at 20° C.

IL-8 Measurement

Two sandwich ELISAs were used for measurement of free and total IL-8. An in-house assay (1) uses a capture antibody (gift of Dr I. Lindley) that detects an epitope of IL-8 that is concealed by macromolecular binding to IgG; hence, only free IL-8 is detected (22). Additionally, under the conditions used in the binding assays described above, DNA and actin at 0.5 mg/ml masked detection of hrIL-8 by 40%.

Total IL-8 was measured by a commercial kit (PeliKine; Eurogenetics, Hampton, UK)

Neutrophil Purification

Neutrophils were purified from EDTA anticoagulated normal venous blood. Red blood cells were removed by sedimentation with 6% (wt/ vol) dextran 70 (Macrodex; Pharmacia) for 45 min at 20° C. Leukocyte-rich supernatants were underlayered with an equal volume of Lymphoprep (Nycomed Pharma, Oslo, Norway) and tubes were centrifuged at 300 × g for 30 min at 20° C. The upper layers were discarded and the remaining red blood cells subjected to hypotonic lysis. Cells were resuspended at 10⁶/ml in HBSS (GIBCO) at pH 7.4 with 20 mM HEPES. Neutrophils were greater than 98% pure.

Neutrophil Chemotaxis Assay

After treatment with buffers or mucolytics as above, sputum supernatants were added to the lower wells of a micro-Boyden chemotaxis chamber. Neutrophils that had been purified from normal human blood were placed in the upper wells of the chamber, which were separated from the lower wells by a porous (5-µm pore size) polyvinylpyrrolidone-free polycarbonate membrane. The chamber was incubated for 30 min at 37° C in a humidity-controlled environment. Migrated cells that were adherent to the lower side of the membrane were fixed in methanol and stained (Hema-Gurr; BDH, Poole, UK) for counting. Cell counts represent the average of the number of cells counted in five fields of view (original magnification: ×400).

Neutralization of IL-8

To confirm the role of IL-8 in the enhanced chemotactic activity, experiments were conducted in the absence and presence of a neutralizing anti-IL-8 antibody. Sputum supernatants were preincubated for 1 h at 4° C with a goat polyclonal anti-IL-8 (20 µg/ml; gift of Dr. I. Lindley) before use in chemotaxis assay.

Neutrophil Binding Assay

To elucidate the mechanism by which DNA and actin inhibit neutrophil chemotaxis, neutrophil receptor binding assays were carried out using ¹²⁵I-labeled IL-8 that had been incubated with DNA, actin, or a buffer control. ¹²⁵I-labeled IL-8 (0.63 nM; Amersham, Buckinghamshire, UK) was incubated either alone or with DNA (5 mg/ml; Sigma) or actin (0.5 mg/ml; Sigma) (dilutions in PBS/1% BSA; final volume, 100 µl) overnight at 20° C. Then 10⁶ neutrophils (purified as above, diluted in 100 µl PBS/1% BSA) were added to each incubation for 1 min at 20° C. The cells were pelleted in a microfuge (2,000 × g for 2 min at 4° C) and washed three times in ice-cold PBS/1% BSA. Finally, the pellets were assayed for  emission (1 min; LKB-Wallac  counter).

Statistical Methods

The effects of mucolytics on sputa were compared with the effects of buffers alone using Wilcoxon's signed ranks test for paired data.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Free IL-8 binds to the heparin column and requires high salt (0.6 M NaCl) elution to remove it (Figure 1A). IL-8 that was incubated with DNA for 4 h continued to bind to the heparin column, suggesting that it was still free and had not bound to the DNA (Figure 1B). After an overnight incubation with DNA, 57% of the IL-8 was prevented from binding to the heparin column and eluted in the wash-through fractions (Figure 1C). Overnight incubation with F-actin resulted in 72% of the IL-8 being prevented from binding to the heparin column (Figure 1D). These results suggest that, in part, IL-8 had become bound to the polyanion in each mixture. We were able to support these findings by separating the same incubation mixtures using gel filtration chromatography. IL-8, incubated overnight with PBS, eluted from the gel filtration column with a molecular weight of 8 kD (Figure 2A). After coincubation with either DNA (Figure 2B) or actin (Figure 2C), IL-8 eluted from a gel filtration column at apparent molecular weights as great as and in excess of 600 kD, indicative of binding to DNA and F-actin.

View larger version (19K): [in this window] [in a new window]   Figure 1.   Heparin affinity chromatography. Elution profile of IL-8 incubated with PBS overnight (A), with DNA for 4 h (B), with DNA overnight (C), or with actin overnight (D), indicated by black bars. The black line indicates the NaCl elution gradient. The graphs are each representative of three separate experiments.

View larger version (13K): [in this window] [in a new window]   Figure 2.   Gel filtration chromatography. (A) IL-8 (50 µg/ml) incubated overnight with PBS. (B) IL-8 incubated overnight with DNA (5 mg/ml). (C) IL-8 incubated overnight with actin (0.5 mg/ml). The graphs are each representative of three separate experiments.

Treatment with gelsolin increased the concentration of free IL-8 in CF sputa from 1.2 ± 0.8 ng/g in the buffer controls to 4.8 ± 2.3 ng/g, whereas DNase treatment increased the concentration from 0.9 ± 0.5 ng/g in the controls to 9.3 ± 2.9 ng/g (Figure 3). The concentration of total IL-8 (97.8 ± 70.5 ng/g) remained the same between buffer and mucolytic-treated samples, as expected. Thus, free IL-8, which represented only approximately 1% of the total IL-8 in buffer-treated sputum, accounted for nearly 10% after treatment with DNase I and 5% with gelsolin.

View larger version (14K): [in this window] [in a new window]   Figure 3.   Free IL-8 in CF sputa before and after treatment with DNase (D), gelsolin (G), or buffer (B). Sputum specimens (n = 10) were treated with DNase (100 µg/ml) or buffer control, or with gelsolin (250 nM) or buffer control. Sputum supernatants were assayed by ELISA for free IL-8. Asterisks indicate p < 0.05 for the comparison of enzyme treatment with buffer control. The values represent means ± SEM.

The number of migrated neutrophils (expressed as neutrophils per high power field, n/hpf) was significantly increased (p = 0.05) in response to sputa treated with DNase I (31.6 [4.8-52.2] n/hpf) compared with its buffer control (17.2 [5.8-44.8] n/hpf) (Figure 4). There was also a significant increase (p = 0.003) in the chemotactic response of neutrophils to gelsolin-treated sputa (182 [59-283] n/hpf) compared with its buffer control (96.4 [27-127] n/hpf). It was also evident that the response to sputa treated with gelsolin buffer, containing 1 mM DTT, was significantly higher than that to the DNase buffer-treated (0.15 M NaCl) samples.

View larger version (13K): [in this window] [in a new window]   Figure 4.   Neutrophil chemotactic activity of CF sputa before and after treatment with DNase (D), gelsolin (G), or buffer (B). Sputum specimens (n = 10) were treated with DNase (100 µg/ml) or with buffer control, or with gelsolin (250 nM) or buffer control. Sputum supernatants were assayed by neutrophil chemotaxis assay. Dots represent individual data points; dashes represent the median value for each treatment.

Addition of the IL-8 neutralizing antibody returned the chemotactic activity of either sputa treated with DNase I or buffer-treated sputa to the baseline level (Figure 5). The results indicate that the increase in chemotactic activity of sputa treated with DNase I was entirely attributable to IL-8. However, in the presence of antibody, gelsolin-treated sputum supernatants retained a small, but significantly increased, chemotactic activity compared with its buffer control.

View larger version (15K): [in this window] [in a new window]   Figure 5.   The IL-8 mediated chemotactic activity of CF sputum after treatment with mucolytics (DNase, D; gelsolin, G) in the absence and presence of an IL-8 neutralizing antibody. Results are expressed as percent response to samples treated with buffer alone. The responses in the absence (light bars) and the presence (dark bars) of the IL-8 neutralizing antibody are shown. Asterisks indicate p < 0.05 for the comparison of samples treated with mucolytic versus buffer control. Dagger indicates that after addition of the IL-8 neutralizing antibody, the response to gelsolin-treated samples remained significantly increased (p < 0.05) compared with buffer- and DNase-treated samples. The assays were performed on sputa from three different patients with CF. The values represent means ± SEM.

Comparison with the binding of radiolabeled IL-8 alone to neutrophils revealed that the addition of DNA resulted in 41 ± 5% inhibition of binding, whereas addition of actin resulted in 67 ± 22% inhibition (Figure 6). These results indicate that DNA and actin inhibit neutrophil chemotaxis by reducing the ability of IL-8 to bind to its specific receptor on the neutrophil.

View larger version (10K): [in this window] [in a new window]   Figure 6.   The binding of 125I labeled IL-8 to neutrophils in the presence and absence of DNA and actin. Asterisks indicate that  counts were significantly (p < 0.05) reduced when the cells had been incubated with radiolabeled IL-8 in the presence of either DNA (5 mg/ml) or actin (0.5 mg/ml). The assays were performed on neutrophils from three separate donors. The values represent means ± SEM.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Taken together, our results overturn the conventional dogma that the anionic polymer DNA can bind cationic enzymes such as neutrophil elastase and cathepsin G, but not the cationic chemokine IL-8 (13). Despite endogenous DNase (7) and gelsolin (23) in the CF airways, treatment of CF sputa with these mucolytics in vitro significantly increased the concentrations of bioactive IL-8. This is further supported by our results demonstrating binding of IL-8 by the polyanions in vitro, and subsequent inhibition of IL-8 function.

Binding of IL-8 to DNA and actin polymers was observed after overnight incubation, resulting in the appearance of complexes with molecular weights in excess of 600 kD on gel filtration chromatography. The reduced binding of IL-8 to DNA at the 4-h time point may reflect the viscous nature of the sample and poor mixing of cytokine and polymer. DNA and actin were used at concentrations known to be present in the CF airways; therefore, these polymers are likely to contribute to the high molecular weight forms of IL-8 that we previously reported (2). In complex with DNA and actin, IL-8 was prevented from binding to a heparin affinity column. Therefore, the affinity of IL-8 for DNA and actin is apparently greater than that for heparin, which binds to IL-8 with an affinity of 10⁶-10⁵ M. Because we have reported undetectable levels of heparin in CF airways (24), DNA and actin may be relatively important in the regulation of IL-8 function.

Previous reports have shown that rhDNase therapy is accompanied by an initial marked increase in neutrophil elastase and cathepsin G activity (13, 17, 18). Measurements of sputum elastase made during these trials appear to depend on initial sputum elastase levels. Patients with mild to moderate disease (13, 17) show a decrease in elastase level with continuing treatment, whereas those with moderate to severe disease (and higher initial elastase levels) (18) showed an increase in elastase level that only declined moderately at the end of the trial. It has been speculated that these observations may account for the failure of many less well patients to benefit from DNase therapy (19). More recently, CF sputum that has been treated in vitro with hrDNase I has been demonstrated to induce lung hemorrhage in mice, a feature ascribed to the release of DNA-bound elastase (25). It is suggested that this mechanism may account for patients with CF who present with hemoptysis while receiving DNase therapy.

The polyanionic nature of DNA and actin dictates that in vivo they are found complexed with cationic peptides, e.g., DNA with histones (8), actin with myosin (26). With no net negative charge, DNA and actin are unlikely to bind IL-8. However, these binding peptides are highly susceptible to proteolysis by elastase (10, 27), a protease that is found in unusually high concentrations in the CF airways (11). Once stripped of their "cationic shield" DNA and actin have the potential to bind further cations, including IL-8 and other chemokines. Although clearly supporting an IL-8 binding role for DNA and actin, our data further suggest that the increase in chemotactic potential of sputa after gelsolin treatment, although largely attributable to IL-8 release, may partly relate to the release of other cationic neutrophil chemoattractants such as ENA-78 (28). Purified DNA and actin, alone, bound and masked detection of free IL-8 in our assay. However, the 4- to 10-fold increase in free IL-8 detected in sputa after depolymerization of actin and DNA may reflect IL-8 released from copolymers of actin and DNA in sputa, the presence of actin leading to the formation of large DNA fibers that do not form with DNA alone (29). The effect of DNase was observed using a concentration of enzyme higher than that achieved in the clinical setting; mean, 2.9 µg/ml (30). This allowed for a direct comparison with a previous report (13) regarding the lack of effect of DNase on IL-8 concentrations in CF sputa. The effect of the gelsolin buffer on neutrophil chemotactic activity in sputa was presumably due to IL-8 released via the mucolytic effect of DTT on mucins, which are also polyanionic molecules contributing to sputum viscosity.

The effects we observe may underlie the observation that DNase is not uniformly beneficial in patients with CF, and that although improvements in lung function are seen in some patients, deterioration is seen in others (20, 31). The concentration of DNA, and presumably actin, in CF sputum increases with severity of disease, bacterial infection, and decreased lung function (5, 32, 33). Similarly, the concentrations of IL-8 and elastase are increased in less well patients (1, 11), suggesting high concentrations of IL-8 in DNA and actin-complexed forms, and a greater risk of increased inflammation by treatment with mucolytics. Results of clinical trials with DNase I suggest that although lung function improves for about 30% of patients (31), at least in the short to medium term, there is no evidence of reduced inflammation (19, 31), for which this report offers one explanation. To date, no predictive markers of the potential benefit of DNase therapy in individual patients with CF have been described. However, a pretreatment assessment of the in vitro generation of free IL-8 by mucolytics might prove useful. Gelsolin has not been shown to improve pulmonary function (34, 35), and additive rhDNase and gelsolin therapy has been suggested (34). Our results indicate that an in vitro investigation of the combined effects of DNase and gelsolin on active inflammatory mediators in CF sputa is warranted.

The presence of complexed forms of IL-8 within the airways has only recently been addressed (2), and it is now necessary to relate form to function. Concentrations of IL-8 in the airways are increased in a number of inflammatory airways diseases, from CF to bronchial asthma and eosinophilic pneumonia (36, 37). A strong negative correlation between the concentration of free IL-8 in the airways of patients with CF and the Shwachman score, a measure of well being, indicates its central role in the pathophysiology of this disease (1). However, free IL-8 in the bloodstream of patients with CF does not correlate with disease progression, suggesting that the inflammatory response is distinctly compartmentalized (1). It has been shown that IL-8 in the bloodstream is almost entirely complexed with a neutralizing IgG autoantibody (22). In the airways of patients with asthma there is an increase in the level of IL-8, but this is largely present in complexes with specific IgA autoantibodies, and levels of free IL-8 are only increased in severe cases of the disease (37).

A failure to recognize IL-8, and presumably other chemokines, in macromolecular forms will generate data open to misinterpretation regarding the mediator function of the chemokine.

In conclusion, we have shown that DNA and actin do bind IL-8. Further, after mucolytic therapy, concentrations of bioactive IL-8, released from macromolecular complexes with polyanions present in the airways, may be increased. Measurements of IL-8 in complex with DNA or actin may therefore prove a useful predictive marker of those patients who will benefit from such therapy. Studies are currently under way to investigate these possibilities.