Apolipoprotein A-I binds and inhibits the human antibacterial/cytotoxic peptide LL-37.

The antibacterial and cytotoxic activity of the human cathelicidin peptide LL-37 is inhibited by plasma. Because LL-37 does not undergo rapid degradation in human plasma, we postulated that this inhibition results from binding of LL-37 to unidentified proteins. An LL-37 binding plasma protein has now been isolated by affinity chromatography. SDS-polyacrylamide gel electrophoresis of proteins that bound to an LL-37 column revealed one band with a molecular mass of about 26 kDa, and amino acid sequence analysis identified the protein as apolipoprotein A-I (apoA-I). Biomolecular interaction analysis using surface plasmon resonance showed that LL-37 and isolated apoA-I bind with an apparent Kd in the low micromolar range. 50 microM of apoA-I inhibits the antibacterial activity of 50 microM LL-37 by about 50% of the inhibition exhibited by plasma. In addition, anti-apoA-I IgG completely blocks the plasma inhibition of LL-37 antibacterial activity up to a peptide concentration of 25 microM and blocks most of the plasma inhibition at higher LL-37 concentrations. These results indicate that apoA-I is the main LL-37 binding protein in human plasma and may work as a scavenger of LL-37, thus suggesting a novel mechanism involved in the regulation of a cathelicidin peptide.

The antibacterial and cytotoxic activity of the human cathelicidin peptide LL-37 is inhibited by plasma. Because LL-37 does not undergo rapid degradation in human plasma, we postulated that this inhibition results from binding of LL-37 to unidentified proteins. An LL-37 binding plasma protein has now been isolated by affinity chromatography. SDS-polyacrylamide gel electrophoresis of proteins that bound to an LL-37 column revealed one band with a molecular mass of about 26 kDa, and amino acid sequence analysis identified the protein as apolipoprotein A-I (apoA-I). Biomolecular interaction analysis using surface plasmon resonance showed that LL-37 and isolated apoA-I bind with an apparent K d in the low micromolar range. 50 M of apoA-I inhibits the antibacterial activity of 50 M LL-37 by about 50% of the inhibition exhibited by plasma. In addition, anti-apoA-I IgG completely blocks the plasma inhibition of LL-37 antibacterial activity up to a peptide concentration of 25 M and blocks most of the plasma inhibition at higher LL-37 concentrations. These results indicate that apoA-I is the main LL-37 binding protein in human plasma and may work as a scavenger of LL-37, thus suggesting a novel mechanism involved in the regulation of a cathelicidin peptide.
Naturally occurring antibacterial peptides are widespread and play an important role in host defense (1). These peptides are derived from gene-encoded precursors by proteolysis. Cathelicidins constitute one family of such peptides that has been identified in bovine, porcine, rabbit, mouse, ovine, and human tissues (2)(3)(4). Cathelicidins have a conserved N-terminal region followed by a highly variable C-terminal antibacterial domain. The sequence of the proregion is similar to cathelin, isolated from porcine leukocytes as a cysteine proteinase inhibitor (5). The C-terminal parts generate the active peptides, which can be divided into three structurally different groups. One group contains ␣-helical peptides; another consists of peptides rich in certain residues, such as Pro, Arg, and Trp; and a third group are peptides with one or two disulfide bonds. Cathelicidins exhibit a broad spectrum of antimicrobial activity, and fungicidal activity (6).
The human cathelicidin LL-37 was originally predicted from a cDNA clone (7), and the mature peptide LL-37 was isolated from degranulated granulocytes (8). LL-37 is active against both Gram-positive and Gram-negative bacteria. The LL-37 CAMP (cathelicidin antimicrobial peptide) gene is up-regulated during inflammatory skin disorders, whereas no expression was detected in normal skin (9). Furthermore, the gene is highly expressed in lung epithelia, and significant expression is also found in other epithelial tissues like the gastrointestinal tract (10), suggesting that LL-37 participates in the human first line of defense. LL-37 is highly cationic and transforms from an unordered to amphipathic ␣-helical structure in an anion-, pH-, and concentration-dependent manner; the antibacterial activity of LL-37 correlates with the ␣-helical content (11).
The minimal inhibitory concentration of LL-37 against Escherichia coli D21 is 5 M, and LL-37 down to 15 M is cytotoxic to eukaryotic cells (11). Therefore it seems likely that released LL-37 is harmful to host cells. However, in human plasma both the antibacterial and cytotoxic activity of LL-37 is inhibited, which is not caused by proteolysis (11). We have now isolated the main LL-37 binding protein from human plasma and identified it as apolipoprotein A-I (apoA-I). 1 Investigations of the interactions between LL-37 and apoA-I suggest that apoA-I can function as a scavenger of LL-37 and thereby prevent host cell damage.

EXPERIMENTAL PROCEDURES
Materials-Streptavidin-agarose, human apoA-I, human albumin, and egg phosphatidylcholine (PC) were from Sigma. Sheep anti-human apoA-I IgG was from The Binding Site. Human plasma was from the Department of Transfusion Medicine at Karolinska Hospital. Phast gels were from Pharmacia, and Sep-Pak cartridges were from Waters. All chemicals for peptide synthesis were from Perkin-Elmer.
Peptide Synthesis-Biotinyl-Lys-(Gly) 3 -LL-37 (biotinylated LL-37) was synthesized with tert-butyloxycarbonyl chemistry in an ABI 430A peptide synthesizer (Perkin-Elmer), starting from 1 g of resin-bound LL-37 (11). First three Gly residues were added, and thereafter N ⑀biotinyl-Lys (Bachem) dissolved in dimethylformamide was added using double coupling. The protecting groups were removed, and the peptide was released from the resin by cleavage in anhydrous hydrogen fluoride/anisol/dimethylsulfide (10:1:1 v/v/v) for 80 min at 0°C. After removal of scavenger and protecting groups with diethylether, the peptide was recovered in 30% acetic acid and lyophilized. Biotinylated LL-37 was isolated by reversed-phase HPLC using a C 18 column and a linear gradient of acetonitrile in aqueous 0.1% trifluoroacetic acid. The mass of the purified peptide determined by matrix-assisted laser desorption/ionization mass spectrometry was 5023 Da (calculated 5019 Da). The antibacterial activity of biotinylated LL-37 is identical to that of LL-37, as determined by the inhibition zone assay (see below). The secondary structure of biotinylated LL-37 in 50 mM sodium phosphate buffer, pH 7.4 or pH 5.0, is indistinguishable from that of LL-37, as judged by CD spectroscopy.
Isolation and Analysis of LL-37 Binding Proteins from Human Plasma-Biotinylated LL-37 (50 nmol) in 1 ml of 0.1 M sodium phosphate buffer, pH 7.4, was mixed with 1 ml of streptavidin-agarose suspension. The mixture was incubated at room temperature for 20 min and then * This work was supported by Swedish Medical Research Council Grants 13X-10371 and 16X-11217, Swedish Cancer Society Grant 1806, Magnus BergvallЈs Foundation, and the Swedish Society of Medicine. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ To whom correspondence should be addressed. Tel.: 46-8-7287781; Fax: 46-8-337462; E-mail: jan.johansson@mbb.ki.se. poured into a glass column (inner dimeter, 0.6 cm). After equilibration with 0.1 M sodium phosphate buffer, pH 7.4, 2 ml of human plasma was added, and the column was incubated for 20 min. Thereafter the plasma was eluted, the column was washed with 10 ml of 0.1 M sodium phosphate, pH 7.4, and finally eluted with 2 ml of 0.1 M sodium phosphate buffer, pH 5.0. The low pH eluate was desalted on a reversed-phase Sep-Pak C 18 column (125 Å pore size) equilibrated with 0.1% trifluoroacetic acid. After washing with 0.1% trifluoroacetic acid, the protein was eluted with 80% acetonitrile/0.1% trifluoroacetic acid. The Sep-Pak eluate was analyzed by SDS-PAGE and sequence analysis. SDS-PAGE was performed in 10 -15% gradient gels with the Phast-system (Pharmacia). Amino acid sequence analysis was carried out in an ABI 494 Protein Sequencer (Perkin-Elmer), and phenylthiohydantoin derivatives were identified by HPLC.
Studies After overnight incubation at 30°C, the diameters of inhibition zones were recorded (12).
Studies of LL-37 and ApoA-I Interactions-Real time interaction studies were performed using Biacore 2000 (Biacore AB) optical biosensor system (13). Biotinylated LL-37 was immobilized on a streptavidin coated SA sensor chip (Biacore AB) to a level of about 1900 response units. A continuous flow (30 l/min) of buffer containing 113 mM NaCl, 24 mM NaHCO 3 , 3.9 mM KCl, 1.3 mM CaCl 2 , 0.6 mM MgCl 2 , and 0.005% Surfactant P20 (Biacore AB), pH 7.3, was maintained over the sensor surface. ApoA-I was diluted in buffer to concentrations from 2.5 to 15 M. Samples of apoA-I (150 l) were injected over immobilized LL-37 at a flow of 30 l/min. After the sample pulse the dissociation of bound apoA-I was recorded for 10 min. Prior to the next sample the LL-37 surface was regenerated with 15 l of 1 M NaCl in 50 mM NaOH. Unspecific binding was monitored by using a reference flow cell containing a streptavidin or biotin-streptavidin-coated sensor chip. There was no difference in the unspecific binding to these two chips. Binding to the reference cell was subtracted from the data generated in the LL-37 cell. All experiments were performed at 25°C. The data were evaluated using Biaevaluation 3.0 software (Biacore AB).

RESULTS
ApoA-I Binds to Immobilized LL-37-Human plasma proteins that bind to biotinylated LL-37 were purified by affinity chromatography. The bound proteins were eluted from the column by 0.1 M sodium phosphate buffer, pH 5.0. pH 5.0 was chosen for elution because at this pH protein side chain carboxylate anions (presumed to be involved in charge interactions with polycationic LL-37) start to get protonated, but LL-37 (and biotinylated LL-37) still exhibits 70% of the helical content observed at pH 7.4. Therefore pH 5.0 buffer was expected to decrease charge interactions between LL-37 and bound protein while retaining a mainly helical conformation of LL-37. SDS-PAGE of the eluate showed only one band with a molecular mass of about 26 kDa (Fig. 1). Edman degradation for 20 cycles unambiguously identified the protein as human apoA-I (Fig. 2). The calculated molecular mass of human apoA-I is 28 kDa and is thus in good agreement with the mass estimated from SDS-PAGE.
The amount of apoA-I that bound to the affinity column was estimated from the inhibition of LL-37 antibacterial activity exhibited by the flow-through, compared with diluted plasma. The flow-through showed the same inhibition as plasma diluted to 85%. Assuming that the decreased inhibition is caused only by removal of apoA-I, this means that 50 M apoA-I (mean apoA-I concentration in human plasma) ϫ 2 ml (plasma volume applied) ϫ 0.15 ϭ 15 nmol of apoA-I bound to the LL-37 column. Provided that there is a 1:1 binding between LL-37 and apoA-I, the maximal binding is 50 nmol. Finally, apoA-I isolated by Sep-Pak chromatography inhibited LL-37 antibacterial activity to the same extent as commercially available apoA-I (see below), and the apoA-I yield during Sep-Pak chromatography was about 75%.
In some affinity purification experiments, albumin was found together with apoA-I. We believe this is caused by the large amounts of albumin in plasma (10 -20 times the apoA-I concentration) and possibly also by secondary interactions between albumin and apoA-I bound to LL-37. No effects of albumin on the LL-37 antibacterial activity could be found (see below).
ApoA-I Inhibits LL-37 Antibacterial Activity-To find out whether apoA-I isolated from human plasma inhibits LL-37 antibacterial activity, LL-37 at different concentrations was mixed with 50 M apoA-I, and the activity against E. coli D21 was analyzed. At 220 M of LL-37, apoA-I exhibited 30% of the inhibitory capacity of plasma, whereas at 50 M of LL-37, 50% of the plasma inhibition was reached, as determined from the inhibition zone diameters, where a doubled LL-37 concentration yields an increase in zone diameter of 1 mm. The discrepancy in inhibitory capacity between isolated apoA-I and plasma could be caused by apoA-I being removed from its plasma environment or by the existence of additional LL-37 binding plasma proteins not detected by affinity chromatography. The following two control experiments indicate that the former alternative (isolated apoA-I being less effective than apoA-I in plasma) is the most likely explanation: (i) Addition of anti-apoA-I IgG to human plasma restores the LL-37 antibacterial activity to that observed in water (Fig. 3). This strongly indicates that no other plasma protein contributes significantly to plasma inhibition of LL-37. The incomplete recovery at high LL-37 Inhibition by ApoA-I 33116 LL-37 concentrations (Fig. 3) may be caused by LL-37 competing with antibodies for apoA-I binding or by LL-37 interacting also with phospholipids (see below). (ii) Albumin at 670 M does not affect LL-37 antibacterial activity, and addition of albumin to apoA-I does not increase the inhibitory capacity. We conclude that apoA-I is the main, or only, LL-37 binding protein in human plasma. However, PC⅐apoA-I complexes as well as PC alone partly inhibit the antibacterial activity of LL-37. It is thus possible that LL-37-phospholipid interactions can contribute to the plasma inhibition of LL-37.
ApoA-I Binds LL-37 with a Low Micromolar Dissociation Constant-Real time interactions between LL-37 and apoA-I were studied by surface plasmon resonance detection using immobilized biotinylated LL-37. Confirming the results from affinity chromatography, biosensor studies showed a binding between LL-37 and apoA-I. An overlay plot of sensorgrams for different concentrations of apoA-I is shown in Fig. 4. The obtained data were used for calculation of apparent association (K a ) and dissociation (K d ) constants. A 1:1 interaction model was found to be the most relevant, as judged from the residuals between fitted curves and experimental data. Global fitting of data from three experiments, covering three to five apoA-I concentrations each, was applied for simultaneous calculation of apparent association and dissociation rate constants. K d for the LL-37-apoA-I interaction was calculated from the kinetic rate constants (k on ϭ 0.7-1.6 ϫ 10 3 M Ϫ1 s Ϫ1 ; k off ϭ 1.0 -2.3 ϫ 10 Ϫ3 s Ϫ1 ) and was in the range 0.6 -2.4 M. K a was in the range 0.4 -1.5 ϫ 10 6 M Ϫ1 . DISCUSSION A protein responsible for plasma inhibition of LL-37 antibacterial/lytic activity was isolated by affinity chromatography and identified as apolipoprotein A-I. ApoA-I is concluded to be the main, or only, plasma protein that binds and inhibits LL-37: (i) isolated apoA-I inhibits the antibacterial activity of LL-37 to 50% compared with the inhibition exerted by whole plasma and (ii) addition of anti-apoA-I IgG to plasma completely blocks inhibition of the antibacterial activity up to 25 M LL-37. In addition, real time binding analysis confirmed the results from affinity chromatography and showed that the LL-37-apoA-I binding has a K d of about 1-2 M.
LL-37 is cytotoxic to human leukocytes at concentrations (15-25 M) that are only three to five times its minimal inhibitory concentration value against bacteria, including Pseudo-monas aeruginosa and E. coli (10,11). In addition, LL-37 lyses human erythrocytes at the cytotoxic concentrations. 2 The LL-37 precursor is present in plasma (14), and the mature peptide is found in bronchoalveolar lavage fluid. 3 This and the fact that other cathelicidins are released from neutrophil granulocytes (15) suggest that LL-37 acts extracellularly. Furthermore, LL-37 is up-regulated during different inflammatory skin disorders, giving a high concentration at inflammatory sites (9). If this applies to inflammatory disorders in general, the existence of mechanisms that protect against LL-37 cytotoxicity appear relevant. LL-37 and apoA-I could interact both at inflammatory sites and in plasma, because the blood vessels are permeable to both proteins and because LL-37 could be released from the precursor in plasma. The apparent K d for the LL-37-apoA-I binding implies that physiological apoA-I plasma concentrations (50 M) will scavenge Ͼ90% of LL-37 at concentrations (25 M) where it is cytotoxic to eukaryotic cells. Thus, our data suggest that apoA-I can function as a scavenger of LL-37 and thereby prevent host cell damage secondary to LL-37 release at inflammatory sites. This is the first report describing inhibitory mechanisms involved in the regulation of cathelicidin antibacterial peptides. However, defensins bind to activated ␣ 2 -macroglobulin in plasma, which was suggested as a scavenging mechanism in inflamed tissues (16).
ApoA-I apparently has several antiinflammatory functions in addition to its possible role in scavenging LL-37; it is involved in the protection against endotoxins (17), binds and inhibits polymerization of complement factor C9 (18), and inhibits IgG-induced neutrophil activation as measured by degranulation and superoxide production (19). The latter function is mediated by free apoA-I but not by high density lipoprotein particles, which may in turn serve as a regulatory mechanism because acute phase proteins may displace apoA-I from high density lipoprotein particles during inflammation (19).
ApoA-I contains amphipathic ␣-helical regions and complement factor C9 and LL-37 likewise contain amphipathic helices. It is tempting to speculate that the amphipathic helices of apoA-I, which are central for its ability to interact with phospholipids, also serve to bind amphipathic peptides. Several cathelicidin peptides are amphipathic and may also be subject to scavenging by apoA-I or other amphipathic plasma proteins. Notably, the antibacterial activity of the porcine cathelicidin PR-39, which is neither lytic nor amphipathic (20,21), is not affected by plasma (11).
In conclusion we show that apolipoprotein A-I binds LL-37, suggesting that this could be a means of attenuating cytotoxic effects at inflammatory sites.