Báo cáo Y học: SMAP-29 has two LPS-binding sites and a central hinge

Eur. J. Biochem. 269, 1181–1189 (2002) (cid:211) FEBS 2002

SMAP-29 has two LPS-binding sites and a central hinge

Brian F. Tack1, Monali V Sawai1, William R. Kearney2, Andrew D. Robertson3, Mark A. Sherman4, Wei Wang5, Teresa Hong5, Lee Ming Boo5, Huiyuan Wu5, Alan J. Waring5,6 and Robert I. Lehrer5,7 Departments of 1Microbiology, 2College of Medicine NMR Facility and 3Biochemistry, University of Iowa, IA, USA; 4Molecular Modeling Core Facility, Beckman Research Institute of the City of Hope, Duarte, CA, USA; Departments of 5Medicine, 6Pediatrics and 7Molecular Biology Institute, UCLA, CA, USA

full-length SMAP-29 showed positive cooperativity, so the EC50 of the peptide (2.6 lM) was considerably lower than that of the individual LPS-binding domains. LPS-binding studies with a mixture of truncated peptides revealed that this cooperativity was primarily intramolecular (i.e. invol- ving the N- and C-terminal LPS-binding sites of the same peptide molecule). CAP-18[106)142], an antimicrobial cath- elicidin peptide of rabbits, resembled SMAP-29 in that it contained N- and C-terminal LPS-binding domains, had an EC50 of 2.5 lM, and bound LPS with positive cooperativity. We conclude that the presence of multiple binding sites that function cooperatively allow peptides such as SMAP-29 and CAP-18 to bind LPS with high a(cid:129)nity.

Keywords: SMAP-29; cathelicidin; LPS; binding; cooper- ativity.

The CD spectra of SMAP-29, an antimicrobial peptide from sheep, showed disordered structure in aqueous buffers, and significant helicity in membrane-like environments, includ- ing SDS micelles, lipopolysaccharide (LPS) dispersions, and trifluoroethanol buffer systems. A structure determined by NMR in 40% perdeuterated trifluoroethanol indicated that residues 8–17 were helical, residues 18–19 formed a hinge, and residues 20–28 formed an ordered, hydrophobic segment. SMAP-29 was flexible in 40% trifluoroethanol, forming two sets of conformers that differed in the relative orientation of the N-terminal domain. We used a chromo- genic Limulus assay to determine the EC50 of the peptide (the concentration that bound 50% of the added LPS). Studies with full-length and truncated SMAP-29 molecules revealed that each end of the holopeptide contained an LPS-binding domain. The higher a(cid:129)nity LPS-binding domain was situated in the flexible N-terminal portion. LPS binding to

(350–600 Da)

trisaccharide

SMAP-29 permeabilizes a variety of cell membranes. It is hemolytic for human erythrocytes [1], renders the inner and outer membranes of Escherichia coli permeable to disaccharide or indicator molecules [2], and rapidly induces a massive potassium efflux from Gram positive and Gram negative bacteria [12]. In this report, we describe the solution structure and LPS binding properties of SMAP-29 and compare them to rabbit and human homologues.

M A T E R I A L S A N D M E T H O D S

Peptide synthesis and purification

SMAP-29, an antimicrobial peptide found in sheep leuko- cytes, possesses potent activity against a broad range of microbial pathogens, including many Pseudomonas aeru- ginosa strains that are highly resistant to conventional antibiotics [1–3]. A homologous peptide, CAP-18[106)142] is present in rabbit leukocytes and has received considerable attention because of its ability to bind LPS [4,5] and its potent antimicrobial activity [3,6,7]. Both SMAP-29 and CAP-18[106)142] are synthesized from an 18-kDa precursor that contains a conserved 11-kDa cathelin domain, and are classified as (cid:212)cathelicidins(cid:213). The only known human cathelicidin is hCAP-18, which carries a 37-residue, largely a helical peptide with antimicrobial [8–10] and LPS-binding [11] activity.

Peptide synthesis reagents, including Fmoc amino acids and coupling solvents, were obtained from PE Biosystems (Foster City, CA, USA) or AnaSpec (San Jose, CA, USA). All organic solvents used for synthesis and purification were HPLC grade or better. The primary structures of the peptides used in this study are found in Table 1. In general, the peptides were made at a 0.25-mmol scale, using FastMocTM chemistry on ABI 431A or 433A peptide synthesizers. We used prederivatized polyethylene glycol-polystyrene (PEG- PS) resins (Perseptive Biosystems, Framingham, MA, USA) and double coupling cycles throughout. The crude product was purified by RP-HPLC on a Vydac C-18 column, using a linear gradient of acetonitrile in dilute (0.1 or 0.085%) trifluoroacetic acid. The molecular mass of the product was confirmed by electrospray mass spectrometry, and its purity was confirmed by analytical HPLC and, in some cases, by capillary electrophoresis.

Correspondence to R. I. Lehrer, Department of Medicine, CHS 37–062, UCLA School of Medicine, 10833 LeConte Avenue, Los Angeles, CA 90095-1690 USA. Fax: + 1 310 206 8766, Tel.: + 1 310 825 5340, E-mail: rlehrer@mednet.ucla.edu Abbreviations: ATR, attenuated reflectance; CFU, colony forming units; Dtrifluoroethanol, perdeuterated trifluoroethanol; EC50, the concentration exhibiting half maximal binding; FTIR, Fourier Transform Infrared; KDO, keto-d-octulosonic acid; LPS, lipopolysaccharide; MEC, minimal effective concentration; MRE, mean residue ellipticity; PGG, a synthetic (cid:212)consensus(cid:213) antimicrobial peptide. (Received 20 July 2001, revised 7 December 2001, accepted 20 December 2001)

1182 B. F. Tack et al. (Eur. J. Biochem. 269)

(cid:211) FEBS 2002

Table 1. Amino-acid sequences of peptides used in this study.

obtain solutions containing 250, 79.1, 25, 7.9, 2.5 or 0.79 lg peptideÆmL)1. An array of 3.2 mm diameter wells was made in the underlay gel, and 8-lL aliquots of the various peptide dilutions were added to them.

Peptide Sequence

PGG GLLRRLRKKIGEIFKKYG SMAP-29[1)29] RGLRRLGRKIAHGVKKYGPTVLRIIRIAG SMAP-29[1)18] RGLRRLGRKIAHGVKKYG SMAP-29[9)29] KIAHGVKKYGPTVLRIIRIAG LGRKIAHGVKKYGPTVLRII SMAP-29[6)25] CAP18[106)142] GLRKRLRKFRNKIKEKLKKIGQKIQGLL PKLAPRTDY

After 3 h, a 10-mL overlay gel [6% (w/v) trypticase soy broth powder, 1% agarose and 10 mM sodium phosphate buffer, pH 7.4] was poured, and the plates were incubated overnight to allow surviving organisms to form microcol- onies. Zone diameters were measured to the nearest 0.1 mm and expressed in units (1 U (cid:136) 0.1 mm), after first subtract- ing the diameter of the well. A linear relationship existed between the zone diameter and the log10 of the peptide concentration. The x-intercept of this line was determined by a least mean squares fit, and this value was considered to represent the minimal effective concentration (MEC).

LPS binding

CD spectrometry

CD measurements were made at 25 (cid:176)C in a 1-mm pathlength cell, using a 62 DS spectropolarimeter (AVIV Associates, Lakewood, NJ, USA), equipped with a thermo- electric temperature controller. The instrument was rou- tinely calibrated with 1 mgÆmL)1 (+)10-camphorsulfonic acid [15]. Mean residue ellipticity (MRE) was expressed as [Q]MRE (degÆcm2Ædmol)1). Samples contained 0.2–2.0 mM peptide in either 50 mM sodium phosphate, pH 6.0, 40% trifluoroethanol or 0.1% ((cid:25) 0.22 mM) lipopolysaccharide, or 20 mM SDS micelles. Spectra were collected at 0.2-nm intervals, with an averaging time of 2 s per datum point. The fractional helical content was estimated from the dichroic minimum at 222 nm, as described by Chen et al. [16].

FTIR measurements

Quantitative chromogenic Limulus amoebocyte assays were performed with a QCL-1000 kit (BioWhittaker, Walkers- ville, MD, USA) as previously described [11]. Briefly, the incubations were performed in flat-bottom, nonpyrogenic 96 well tissue culture plates (Catalog no. 3596, Costar(cid:210), Cambridge, MA, USA). Stock solutions of polymyxin B (Sigma; 7600 UÆmg)1), SMAP-29 and truncated variants were prepared in endotoxin-free acidified water (0.01% acetic acid) and serially diluted in this solution. First, 25 lL of peptide solution and 25 lL of a 1-endotoxin-unit per mL solution of E. coli 0111:B4 lipopolysaccharide were mixed and incubated for 30 min at 37 (cid:176)C to permit peptide–LPS binding to reach equilibrium. Then, 50 lL of the amoebo- cyte lysate reagent was added and exactly 10 min later, 100 lL of chromogenic substrate (Ac-Ile-Glu-Ala-Arg- p-nitroanilide) was introduced. Thereafter, the incubation continued at 37 (cid:176)C for 20 min while liberation of p-nitro- aniline was monitored every 60 s at 405 nm, with a SpectraMax 250 Kinetic Microplate Spectrophotometer (Molecular Devices, Sunnyvale, CA, USA). The DD between 10 and 16 min was calculated for the control sample (containing peptide but no LPS), and from this value the DD between 10 and 16 min for the experimental samples, which contained peptide plus LPS, was subtracted. The percentage reduction in procoagulant activation was directly proportional to the percentage of LPS bound.

Infrared spectra of SMAP-29 were recorded using a Bruker Vector 22TM FTIR spectrometer equipped with a deuterated triglycine sulfate detector. Solvent spectra were obtained by subtracting the deuterated solvent spectrum from the SMAP-29/trifluoroethanol buffer (40 : 60, v/v) solution. Spectra of SMAP-29 in LPS micelles were recorded after drying the dispersion onto a germanium attenuated reflect- ance (ATR) crystal and hydrating the substrate and sample with D2O for 2 h in a Pike horizontal ATR accessory (Pike Technologies, Madison, WI, USA). Spectra encompassed 32 scans taken at a gain of four and a resolution of 1 cm)1.

NMR

Hill plots [13] were performed by plotting log10 peptide or lipopeptide (polymyxin B) concentrations against log10 [(FI)/(1.0 ) FI)], where FI was the fractional inhibition of procoagulant activity observed in the chromogenic assay. Thus, a FI of 0.75 would correspond to a 75% inhibition of procoagulant activity.

Radial diffusion assays

Purified peptides were serially diluted with acidified water (0.01% acetic acid) containing 0.1% human serum albumin (Sigma), as described previously [11,14]. The test bacteria, E. coli strains ML-35p, DH5a and ATCC 33780, were grown to mid-logarithmic phase in trypticase soy broth and washed. Approximately 4 · 105 colony forming units (c.f.u.) per mL were incorporated into a thin (1.2 mm) underlay gel that contained 1% (w/v) agarose (Sigma) in 10 mM sodium phosphate buffer, pH 7.4, with 0.3 mgÆmL)1 trypticase soy broth powder and 100 mM NaCl. Peptides were serially diluted in acidified water with albumin to

For structure determination, the 0.7-mL samples contained 1 mM SMAP-29 and 50 mM phosphate, pH 5.94 in either 60% D2O (Isotec (cid:212)100%(cid:213))/40% perdeuterated trifluoroeth- anol (Dtrifluoroethanol, Cambridge Isotope Laboratories) or 60% H2O/40% Dtrifluoroethanol, and were placed in Kontes Model 240 NMR tubes. A standard set of DQFCOSY, TOCSY and NOESY spectra were collected at 25 (cid:176)C on a Varian 500 MHz instrument (University of Iowa College of Medicine NMR Facility, IA, USA). NOESY spectra were also collected at 10 (cid:176)C and 4 (cid:176)C, to assess temperature dependent changes in peptide confor- mation. All spectra were taken with a 6000-Hz spectral width, solvent suppression by presaturation during a 2.5-s relaxation delay, and States–Haberkorn phase-sensitive

LRKFRNKIKEKLKKIG CAP18[106)126] GLRKRLRKFRNKIKEKLKKIG CAP18[109)126] KRLRKFRNKIKEKLKKIG CAP18[111)126]

SMAP-29: conformation and LPS binding (Eur. J. Biochem. 269) 1183

(cid:211) FEBS 2002

Table 2. Classification of NOE distance constraints used in SMAP-29 structure.

detection in the f-1 dimension. High power 90(cid:176) pulse widths were 7 ls in all spectra. DQFCOSY spectra had a resolution of 2048 complex points in the f-2 dimension and 600 complex points in f-1. Thirty-two transients were averaged for each increment. The resolution of TOCSY and NOESY spectra were 1024 complex points in f-2 and 512 complex points in f-1. In these spectra, 16 transients were averaged per f-1 increment. TOCSY spectra were collected with mixing times of 30, 80 and 120 ms and a spin-lock field strength of 8000 Hz. NOESY spectra were obtained with mixing times of 50, 100, 150 and 300 ms. Spectra were referenced to the Dtrifluoroethanol signal at 3.88 p.p.m.

Constraint type Number of constraints

Samples for hydrogen exchange studies were prepared by subjecting 0.7 mL of 1 mM SMAP-29 solution in 50 mM phosphate buffer (pH 5.94) to lyophilization. To begin the exchange study, 0.7 mL of a 60% D2O/40% Dtrifluoro- ethanol solution was added to the lyophilate. The sample was then immediately capped, inverted four times to mix it, transferred to an NMR tube, and placed in a pre-shimmed magnet at 25 (cid:176)C. Acquisition of spectra commenced immediately after insertion, (cid:25) 90 s after mixing began. The first spectrum was complete 51 s after insertion, and new spectra were taken at 5-min intervals for 1.5 h. Each spectrum resulted from accumulation of 16 transients containing 4096 complex data points, with a 2.5-s relaxation delay and 90(cid:176) pulse widths. In the exchange experiment, the spectral width was 6000 Hz and no presaturation was used.

Processing of NMR data

angle constraints derived from the NMR data. The three hundred initial structures were generated with torsion angles chosen at random. NOE constraints were added to the standard force constant set using the DYANA target function multiplier of u` 0 (cid:136) 10 kJÆmol)1ÆA˚ )2, with all other values as DYANA defaults. Each structure was then annealed in 4000 steps from an initial temperature of 8000 to 0 K followed by 1000 further steps of conjugate gradient minimization. Table 2 details the classification of NOE distance con- straints used to calculate the SMAP-29 solution structure. Lipophilic potentials [18] were calculated for the lowest energy conformer in each structural subset. The lipophilic potential was mapped onto a Connoly style, solvent- accessible surface generated with a 1.4-A˚ solvent radius. This portion of the molecular modeling was performed using the SYBYL package (Tripos, Inc., St Louis, MO, USA) on a Silicon Graphics Octane (R10000) workstation in the College of Medicine NMR Facility at the University of Iowa, IA, USA.

The geometry of the final refined conformer sets was evaluated by PROCHECK_NMR [19]. The coordinates for the 40 lowest energy structures of SMAP-29 in trifluoroethanol buffer, together with a full list of restraints have been deposited in the Protein Data Bank (accession no. 1FRY). Modeling was performed on a Silicon graphics Indigo- 2R10000 High Impact workstation (Beckman Research Institute, City of Hope Core Facility), using INSIGHT/ DISCOVER 97.0 software within the DISCOVER environment (Molecular Simulations, San Diego, CA, USA). A ribbon representation for the two families of SMAP-29 structures was constructed from the simulated annealing and geometry optimization studies for each conformer-set of coordinates using INSIGHT software.

Spectra were processed using the VNMR 6.1B software package. The apodization for DQFCOSY spectra consisted of a 0.158-s Gaussian function, shifted by 30(cid:176), in f-2 and an unshifted 0.05-s Gaussian function in f-1. NOESY and TOCSY spectra were apodized with a 0.079-s Gaussian function, shifted by 30(cid:176), in f-2 dimension and by an unshifted Gaussian of 0.04 s in the f-1 dimension. NOESY spectra were baseline corrected prior to peak volume measurements. If needed, a low pass digital filter was applied on transformation to remove any remaining solvent signal. Coupling constants were extracted from the DQFCOSY spectra by modeling line shapes of aH-NH- cross peaks. Spectra in the exchange experiments were apodized by a 0.73-Hz Lorentzian line broadening prior to Fourier transformation. The spectra were then baseline- corrected using a cubic spline function. Amide peak intensities as a function of time were measured using the VNMR 6.1B software. Intensity vs. time profiles for peaks with measurably slow exchange rates were then fitted to exponential functions using VNMR 6.1B.

Intraresidue Interresidue One residue away Two residues away Three residues away Four residues away < 2.50 A˚ > 2.50 and < 3.50 A˚ > 3.50 A˚ 64 106 70 9 18 9 42 58 70

R E S U L T S

Calculation of structures and molecular modeling

LPS Binding

Figure 1 shows that full length SMAP-29 bound E. coli LPS with an apparent affinity constant (EC50) of 2.6 lM. SMAP-29 [6–25], a 20-mer that lacked residues 1–5 and 26–29 of the holopeptide bound LPS weakly, with an estimated EC50 of (cid:25) 0.7–1.0 mm. As PGG, an 18 residue peptide (Table 1) that we used as a control, had an EC50 of 7.1 lm, the inability of SMAP-29 [6–25] to bind LPS was not simply attributable to the shorter length. Because SMAP-29 [1–18] and SMAP-29 [9–29] had EC50 values of 49 lM and 143 lM, respectively, the presence of at least two

Peak volumes from the VNMR package were converted to XEASY format using the program VNMR2XEASY (W. R. Kearney, unpublished work, available on request). aH-NH coupling constants and NOESY peak volumes were then used to generate geometric constraints using the DYANA package [17]. Examination of representative build up curves showed no evidence of spin-diffusion, so peak volumes were extracted from the 300 ms mixing time spectrum. In DYANA, a total of 300 trial structures were created and annealed using 170 distance constraints and eight aH-NH torsion

1184 B. F. Tack et al. (Eur. J. Biochem. 269)

(cid:211) FEBS 2002

coefficient of the mixture was 1.86, which, together with its EC50, suggests that while some intermolecular cooperativity exists, its contribution to overall LPS binding by SMAP-29 was relatively minor.

We also used the Limulus chromogenic assay technique to examine LPS binding by several other peptides, including CAP-18[106)142], the 37-residue homologue of SMAP-29 in the rabbit. The results can be seen in Fig. 3. The CAP-18[106)142] binding isotherm was also sigmoidal and had an EC50 of 2.6 lM. When these data were graphed on a Hill plot, the coefficient was 2.78 (cid:139) 0.38 (mean (cid:139) SEM) (r) was 0.964. and the line’s correlation coefficient CAP-18[106)126] a 21-mer containing only the N-terminal portion of the holopeptide, retained considerable affinity for LPS (EC50, 12.1 lM), but its isotherm was not sigmoidal and the Hill coefficient of 0.78 (cid:139) 0.065 (mean (cid:139) SE), signified that binding events were non-cooperative.

We also studied CAP-18[106)126], an 18-mer that was identical to the 21-mer except for lacking its first three N-terminal residues (GLR). This peptide showed a >12-fold reduction in affinity for LPS (EC50 (cid:136) 151 lM), implicat- ing these initial residues in the LPS binding site. Further N-terminal truncation to produce a 16-mer, CAP- 18[106)126], virtually abolished LPS binding, confirming the existence of this N-terminal LPS binding site. In an earlier study, we examined the affinity of the human cathelicidin, LL 37 for E. coli LPS. The EC50 value was 0.36 lM [13] and the Hill coefficient of 2.02 was consistent with positive cooperativity.

LPS binding sites was suggested, one at each end of the molecule. The threefold greater affinity of SMAP-29 [1–18] for LPS relative to SMAP-29 [9–29] suggests that the N- terminal binding site has higher affinity for LPS than does the C-terminal one. The presence of two binding sites and the sevenfold to 21-fold greater affinity of the holopeptide for LPS, indicates that binding by the holopeptide is cooperative. The sigmoidal shape of the SMAP-29 binding isotherm shown in Fig. 1 also indicates cooperativity, as does the coefficient of 2.66 (cid:139) 0.19 when binding is graphed on a Hill plot (Fig. 2).

Correlation between LPS binding and antimicrobial potency

Figure 4 compares the previous EC50 data with the minimal effective concentrations (MECs) of PGG, SMAP-29 and the truncated SMAP-29 variants for three different strains of E. coli. The two properties were correlated, as a higher affinity for E. coli LPS was associated with a lower MEC for the organism in the presence of approximately physio- logical (100 mM) concentrations of NaCl.

CD Measurements

To determine whether the cooperativity was intermole- cular or intramolecular, we measured LPS-binding by an equimolar mixture of SMAP-29 [1–18] and SMAP-29 [9–29] (Fig. 2). One of these peptides (SMAP-29 [1–18]) lacks the C-terminal LPS-binding site while the other lacks the N-terminal site, therefore we anticipated that if the cooper- ativity of SMAP-29 was primarily intermolecular (i.e. involving LPS binding sites on different peptide molecules), the mixture would simulate full length SMAP-29 in its binding. The mixture had an EC50 of 27.8 lM, somewhat better than SMAP-29 [1–18] (EC50 of 49 lM), but 10-fold higher than SMAP-29 (EC50 of 2.6 lM). We therefore infer that cooperativity in LPS-binding is primarily intramolecu- lar, whereas binding of one LPS-binding site in a SMAP-29 molecule facilitates binding by the other site. The Hill

In aqueous phosphate buffer, SMAP-29 had a mostly disordered conformation (Fig. 5). However, when added to LPS dispersions or to the micelles formed by 20 mM SDS in 50 mM phosphate buffer at pH 6.0 (data not shown), a

Fig. 1. LPS binding by SMAP-29 peptides. The EC50 values derived from these binding isotherms were as follows: full-length SMAP-29, 2.56 lM; PGG, 7.1 lM; SMAP-29 [1–18], 49 lM; SMAP-29 [9–29], 143 lM; SMAP-29 [6–25], (cid:25) 700 lM. The previously reported EC50 values for polymyxin and LL-37 in this assay were (cid:25) 30 and 360 nM, respectively [11].

Fig. 2. LPS binding by a mixture of truncated SMAP-29 peptides. The left panel shows binding isotherms for SMAP-29 (d) and an equimolar mixture of SMAP-29 [1–18] and SMAP-29 [9–29] (s). The right panel shows a Hill plot of these data, with the Hill coe(cid:129)cients adjacent to the lines.

SMAP-29: conformation and LPS binding (Eur. J. Biochem. 269) 1185

(cid:211) FEBS 2002

Fig. 3. LPS binding by CAP18 peptides. The primary sequences of these peptides are shown in Table 1. The EC50 values derived from these binding isotherms were as follows: CAP18[106)142], 2.5 lM; CAP18[106)126], 12.1 lM; CAP-18[109)126], 151 lM; CAP18[111)126], > 500 lM.

generation of structures. The NOE contacts, angular constraints and a-proton chemical shift index together define the structure and are displayed in Fig. 6. No new NH-NH or aH-NH peaks appeared in the NOESY spectra at 10 (cid:176)C to aid in the conformational analysis. Peptide aggregation occurred at 4 (cid:176)C, evidenced by significant broadening of spectral lines. Only Lys9 and Arg26 showed any measurable protection from NH exchange.

Of

predominantly helical conformation was evident. In 40% trifluoroethanol, an organic solvent whose dielectric con- stant resembles those of biological membranes [20], the CD spectrum of SMAP-29 was consistent with (cid:25) 57.2% a helical content (Fig. 5). Similar helicity was observed in SDS (57.0% helix), and LPS (63.2% helix) micelles. At 25 (cid:176)C in 40% trifluoroethanol, the apparent helicity of 0.2–2 mM SMAP-29 varied only from 57.0% to 59.5%, suggesting that it did not aggregate under these conditions. Figure 5B shows that the CD spectra for SMAP-29 [1–18] and SMAP-29 [9–29] in LPS micelles were indistinguishable from each other and from SMAP-29, indicating that N or C-terminal truncation of SMAP-29 did not alter its shape drastically.

FTIR studies in trifluoroethanol buffer and LPS

the 300 conformers calculated from the NMR constraints in Dynamics Algorithm for NMB Applica- tions (DYANA) [17] the 192 lowest energy conformers were retained for further structural analysis. These 192 structures partitioned into two sets, differing by a 180(cid:176) flip in the / angle of Gly18. The results of analyzing the structures are summarized in Tables 2 and 3. The root- mean-square deviation (rmsd) of the backbone atoms between residues 2 and 28 was calculated to be 2.72 A˚ for the 98 conformers in the subset with the Gly18 / (cid:136) 120(cid:176) (designated the L subset). For the 94 conformers with the Gly18 / (cid:136) )60(cid:176) (designated the R subset), this global backbone atom rmsd was 2.94 A˚ . This indicates a high degree of conformational disorder in the structures.

The helical structure of SMAP-29 in trifluoroethanol buffer and LPS micelles was further characterized using FTIR. In the trifluoroethanol buffer system SMAP-29 had a domi- nant absorption at 1647 cm)1 (Fig. 5A), typical of a more nascent helical structure [21]. When SMAP-29 was incor- porated into LPS micelles, the helical absorption peak shifted from 1647 cm)1 to 1655 cm)1 more typical of a highly stabilized helical structure [21].

NMR structure determination in 40% trifluoroethanol

When the sets were analyzed in segments, the presence of highly ordered local secondary structure became apparent. The structure was divided into three segments: a highly flexible region from the N-terminus through a from Arg8 (cid:212)hinge(cid:213) at Gly7, a nearly helical segment through Tyr17, and another nearly helical region follow- ing the (cid:212)hinge(cid:213) at Gly18/Pro19, from Thr20 through Ala28. The backbone rmsd calculated for the whole set of 192 structures over the segment containing residues 2–6 was 1.27 A˚ . The flexibility of this segment is apparent, as shown by the wide distribution of backbone torsion angles for these residues (Table 3). This is also consistent with the lack of long range inter-residue NOE contacts along the backbone of this region (Fig. 6). From residues 8–17, the full set backbone rmsd is 0.72 A˚ (Table 3), the

Complete sequential assignment of the two-dimensional 1H-NMR spectra for SMAP-29 was obtained using the Wuthrich strategy [22]. The spin systems were identified using DQF-COSY spectra, complemented by TOCSY spectra. Partial or complete assignments were made for all residues except Arg1. In those spin systems where complete assignments were not possible due to spectral overlap, backbone resonances were assigned. A total of 170 NOE derived distance constraints were used to generate the structure families, of which 106 were interresidue contacts and 64 were intraresidue contacts. Spectral overlap caused rejection of a further 60 intraresidue contacts from use in the

Fig. 4. LPS binding and antimicrobial potency. The EC50 for binding to E. coli 0111:B4 LPS and the minimal effective concentrations (MEC) for E. coli appear to be correlated.

1186 B. F. Tack et al. (Eur. J. Biochem. 269)

(cid:211) FEBS 2002

Fig. 5. CD spectra of SMAP-29. The spectra shown in the main figure were taken in 50 mM phosphate buffer (d), pH 6.0; 40% trifluoroethanol in 50 mM phosphate buffer, pH 6.0 (s); LPS micelles in 50 mM phosphate buffer, pH 6.0 (- - - -) The concentration of peptide was 0.2 mM, the temperature was 25 (cid:176)C, and the cell path-length was 0.01 cm. Inset (a) shows FTIR spectra of SMAP-29. One spectrum (ÆÆÆÆÆ) was taken in 40% trifluoroethanol in 50 mM phosphate buffer pH 6.0. The other (–––) was taken from a film containing LPS and SMAP-29 at a 10 : 1 molar ratio. Inset (b) shows CD spectra of SMAP-29 [1–18] and SMAP-29 [9–29] in LPS micelles in 50 mM phosphate buffer, pH 6.0. Inset B shows the CD spectra of SMAP-29 [1–18] and SMAP-29 [9–29], taken in LPS micelles in 50 mM phosphate buffer, pH 6.0.

are observed between Leu22 and Arg26 and between Arg23 and Arg26. Arg26 NH displayed some protection from H/D exchange (k (cid:136) 0.00124 (cid:139) 0.00005 s)1), proba- bly due to a hydrogen bond with Leu22 O. The NOE contacts, negative a-proton-chemical shift index (Fig. 6) and NH exchange protection are consistent with consid- erable helicity in this region. Representations of the backbone and important side chains of the lowest energy conformers of the L and R sets are presented in Fig. 7A,B.

backbone torsion angle distributions are narrow, and a large number of NOE contacts of the type expected in a helix (Fig. 6) all indicate that this region is highly ordered and helical. This is confirmed by the negative chemical shift index in residues 7–12 (Fig. 6). Additional support for the helicity of this region is the protection of Lys9 NH from H/D exchange (k (cid:136) 0.0086 (cid:139) 0.0002 s)1), probably indicating a hydrogen bond to Gly7 O. In the region of residues 20–29, the backbone rmsd is 0.57 A˚ , indicating that this region is also highly ordered. Ha-HN contacts

Fig. 6. Summary of the NMR constraints found for SMAP-29. (A) Distance and torsion angle constraints reflect the secondary structure of SMAP-29. Line thickness is proportional to the strength of the NOE. Symbols represent secondary structure as follows: m, helix; ., beta strand; w, helical and beta strand conformations present in ensemble; d, conformations are not within helical or sheet limits. (B) The a-proton chemical shift index [36] is consistent with a helical conformation in all but six residues of SMAP-29.

Table 3. RMSD for backbone atom displacements (A˚ ) and backbone angle displacements ((cid:176)) for global and segmental structural superpositions.

Residue Range RMSD(full) RMSD(L) RMSD(R) RMSD(w) RMSD(/)

2–28 2–6 8–17 20–29 3.44 1.27 0.72 0.57 2.72 1.26 0.71 0.54 2.94 1.27 0.72 0.54 29.65 60.5 18.02 26.4 39.28 51.58 30.75 33.01

SMAP-29: conformation and LPS binding (Eur. J. Biochem. 269) 1187

(cid:211) FEBS 2002

Fig. 7. Models of SMAP-29 in trifluoroethanol buffer. (A) Eight superimposed dynamic models of SMAP-29, four in each conformation (A, black backbone and B, gray backbone), calculated from simulated annealing and geometry optimization. (B) A ribbon representation of SMAP-29 in conformation A (left side) and another in conformation B (right side). In each figure, the flexible N-terminal segment is at the bottom.

D I S C U S S I O N

order, contribute to destabilization, and facilitate peptide insertion and penetration into the bilayer.

We identified a second LPS binding domain of SMAP-29 among its C-terminal residues, VLRIIRIA. Our NMR studies showed this region to be ordered and helical, with one full turn of the a helix in residues 22–26. Whereas the arginines in the VLRIIRIA segment should promote interactions with anionic moieties of LPS, the five apolar residues may interact preferentially with LPS acyl chains and facilitate the peptide’s insertion into the outer mem- brane bilayer.

One aim of this study was to define the LPS-binding domains of SMAP-29. Accordingly, we will begin by briefly describing the structure of LPS. The conserved lipid A domain of LPS forms most of the outer membrane bilayer of Gram-negative bacteria, and is a mono- or di-phospho- rylated b1 fi 6-linked glucosamine disaccharide to which six or seven fatty acids are attached by amide or ester bonds. The inner core oligosaccharide region of LPS contains two or three keto-d-octulosonic acid (KDO) molecules that are linked to two heptose residues that may contain phosphate or other substituents. The minimal LPS structure required for viability in Gram negative bacteria is Re LPS, which consists of lipid A, KDO and heptose residues. Re LPS is tetra-anionic at pH 7, with two negative charges derived from the diglucosamine phosphates and two from the KDO carboxylate anions [23].

We found that SMAP-29, a potently antimicrobial peptide of sheep leukocytes, contained two distinct binding sites for E. coli LPS, one at each end of the molecule. Studies with a mixture of truncated SMAP-29 peptides indicated that these sites bound LPS cooperatively, such that when one terminal domain bound LPS, this event greatly facilitated binding by the other terminal domain.

In addition to testing SMAP-29 and related peptides, we also examined its rabbit homologue, CAP-18[106)142]. We found that the full length peptides, CAP-18[106)142], and ovine SMAP-29 bound E. coli 0111:B4 LPS with almost identical EC50 values, approximately 2.5 lM. Both peptides had sigmoidal binding isotherms and had Hill coeffi- cients between 2 and 3, indicating positive cooperativity. CAP-18[106)126], a 21-mer lacking the 16 C-terminal residues of the holopeptide, retained moderate affinity for LPS (EC50, 12.1 lM), indicating that at least two LPS-binding domains were present. The removal of only three N-termi- nal residues (GLR) from the 21-mer caused a > 12-fold reduction in affinity for LPS (EC50 (cid:136) 151 lM), and remov- ing two additional residues (KR) all but abolished LPS binding. Consequently, the first five residues (GLRKR) of CAP-18[106)142] form an essential part of its high-affinity binding site for LPS. Their similarity to the N-terminal residues of SMAP-29 (RGLRR) is apparent.

In earlier studies that examined binding of LL-37 to LPS, we obtained EC50 values of 0.36 lM for LL-37, along with evidence for positive cooperativity [11]. The LPS binding domains of LL-37 are not yet reported, but its N-terminal residues (LLGDFF) are different in character from those of SMAP-29 or CAP-18. It appears therefore that the a helical antimicrobial peptides of mammalian leukocytes can vary considerably in their structural and segmental organization. Although we did not study CAP-18 peptides by NMR, a published report describes studies with CAP-18[106)137], a slightly truncated variant of CAP-18[106)142] [27]. This peptide had a disordered structure in aqueous media, and formed an unusually stable and rigid alpha helix in 30% trifluoroethanol. Thus, conformational flexibility, while a property of SMAP-29, is not a general property of such peptides. Whether the flexible LPS-binding domain con- tributes to the function of SMAP-29 remains to be determined.

By studying truncated variants of SMAP-29, we localized the highest affinity binding site to eight N-terminal residues, RGLRRLGR. Our NMR studies revealed that these residues were in a relatively flexible region. As four of the eight residues in this N-terminal region are arginines, this portion of the peptide most likely binds anionic groups present in lipid A or inner core polysaccharide regions of LPS. Multiple arginine or lysine residues also constitute a structural motif for LPS recognition by Bactericidal/ Permeability Increasing factor (BP/I), lactoferrin, lysozyme and the antibacterial Limulus anti-LPS factor (LALF) [24]. LPS is normally stabilized by the presence of divalent cations, and it is destabilized when these cations are displaced by EDTA or similar agents. The apparent dissociation constant for divalent cations bound to KDO was reported to be 14 lM [25]. As the EC50 concentrations of SMAP-29 and rabbit CAP-18 for binding E. coli LPS were (cid:25) 2.6 lM, both peptides should be able to displace divalent cations from KDO sites. As the binding of divalent cations to inner core carboxylate anions is believed to promote LPS packing [26], displacement of these divalent cations by peptides could increase outer membrane dis-

1188 B. F. Tack et al. (Eur. J. Biochem. 269)

(cid:211) FEBS 2002

R E F E R E N C E S

1. Skerlavaj, B., Benincasa, M., Risso, A., Zanetti, M. & Gennaro, R. (1999) SMAP-29: a potent antibacterial and antifungal peptide from sheep leukocytes. FEBS Lett. 463, 58–62.

LPS molecules normally occupy up to 80% of the outer leaflet of the outer membrane and are linked by divalent cations to form an oriented and highly ordered structure that provides a formidable barrier to the uptake of exogenous peptides. Initial binding of the arginine-rich N-terminal domains of SMAP-29 or CAP-18 to LPS molecules on the bacterial surface is likely to involve displacement of divalent cations, causing increased mobility and disordered packing of the LPS molecules and their acyl chains. Because SMAP-29 and CAP-18 are considerably larger than the divalent cations they replace, their insertion into the outer leaflet would cause the outer membrane to expand, as was recently demonstrated for CAP-18[106)142] [28]. The resulting architectural chaos should provide the amphipathic C-terminal domain of SMAP-29 even greater access to acyl chains in both leaflets of the outer membrane. This mechanism is consistent with the positive cooperativity we noted in our LPS binding studies, and with the general formulations of (cid:212)self-promoted uptake(cid:213) as shown primarily by the work of Hancock [29,30].

2. Mahoney, M.M., Lee, A.Y., Brezinski-Caliguri, D.J. & Huttner, K.M. (1995) Molecular analysis of the sheep cathelin family reveals a novel antimicrobial peptide. FEBS Lett. 377, 519–522. 3. Travis, S.M., Anderson, N.N., Forsyth, W.R., Espiritu, C., Conway, B.D., Greenberg, E.P., McCray Jr, P.B., Lehrer, R.I., Welsh, M.J. & Tack, B.F. (2000) Bactericidal activity of mammalian cathelicidin-derived peptides. Infect. Immun. 68, 2748–2755.

4. Sawa, T., Kurahashi, K., Ohara, M., Gropper, M.A., Doshi, V., Larrick, J.W. & Wiener-Kronish, J.P. (1998) Evaluation of antimicrobial and lipopolysaccharide-neutralizing effects of a synthetic CAP18 fragment against Pseudomonas aeruginosa in a mouse model. Antimicrob. Agents Chemother. 42, 3269–3275. 5. Larrick, J.W., Morgan, J.G., Palings, I., Hirata, M. & Yen, M.H. (1991) Complementary DNA sequence of rabbit CAP18 – a unique lipopolysaccharide binding protein. Biochem. Biophys. Res. Commun. 179, 170–175.

6. Mason, D.J., Dybowski, R., Larrick, J.W. & Gant, V.A. (1997) leukocyte CAP18 (106–137). Antimicrobial action of rabbit Antimicrob. Agents Chemother. 41, 624–629.

7. Tossi, A., Scocchi, M., Skerlavaj, B. & Gennaro, R. (1994) Identification and characterization of a primary antibacterial domain in CAP18, a lipopolysaccharide binding protein from rabbit leukocytes. FEBS Lett. 339, 108–112.

8. Bals, R., Weiner, D.J., Moscioni, A.D., Meegalla, R.L. & Wilson, J.M. (1999) Augmentation of innate host defense by expression of a cathelicidin antimicrobial peptide. Infect. Immun. 67, 6084–6089. 9. Johansson, J., Gudmundsson, G.H., Rottenberg, M.E., Berndt, K.D. & Agerberth, B. (1998) Conformation-dependent antibac- terial activity of the naturally occurring human peptide LL-37. J. Biol. Chem. 273, 3718–3724.

10. Agerberth, B., Gunne, H., Odeberg, J., Kogner, P., Boman, H.G. & Gudmundsson, G.H. (1995) FALL-39, a putative human peptide antibiotic, is cysteine-free and expressed in bone marrow and testis. Proc. Natl Acad. Sci. USA 92, 195–199.

11. Turner, J., Cho, Y., Dinh, N.N., Waring, A.J. & Lehrer, R.I. (1998) Activities of LL-37, a cathelin-associated antimicrobial peptide of human neutrophils. Antimicrob. Agents Chemother. 42, 2206–2214.

In addition to a helical peptides such as those described here, several other molecules that bind LPS exist, including many with beta-sheet structures [31–33]. Recent studies with horseshoe crab Factor C, a molecule whose autocat- alytic activation by femtogram concentrations of LPS triggers hemolymph coagulation, are of considerable interest. Factor C is a large, multidomain protein, with a C-terminal serine protease domain and N-terminal LPS- binding domains. The N-terminal fragment of Factor C bound LPS with positive cooperativity (Hill coeffi- cient (cid:136) 2.2) [34], and contained three 3.5–4 kDa small consensus repeat (S) domains. Two of these, S1 and S3, bound LPS with EC50 values of 2.25 lM (S1) and 1 lM (S3), quite similar to those of the peptides we studied [35]. Binding isotherms for the S1 fragment were sigmoidal, and the Hill coefficient of 2.42 indicated positive cooper- ativity. The fragment bound to lipid A with positive cooperativity (Hill coefficient (cid:136) 2.2) and contained two LPS-binding short consensus repeat (sushi) domains, S1 and S3, each containing a 34-residue LPS-binding site. The authors suggested that two factors were critical for making Factor C acutely sensitive to LPS; the presence of multiple LPS-binding sites on a single Factor C molecule and their high positive cooperativity in LPS binding. Although the affinity of SMAP-29 or CAP-18 for LPS we measured in our Limulus chromogenic assays was considerably lower than that reported for Factor C, these peptides also contained multiple binding sites and its LPS binding manifested positive cooperativity.

12. Orlov, D., Nguyen, T. & Lehrer, R.I. (2002) Potassium release, a useful tool for studying antimicrobial peptides. J. Microbiol. Methods, in press.

13. Hill, A.V. (1910) The combinations of hemoglobin with oxygen and with carbon monoxide. I. J. Physiol. (London) 40, 4–8. 14. Lehrer, R.I., Rosenman, M., Harwig, S.S., Jackson, R. & Eisen- hauer, P. (1991) Ultrasensitive assays for endogenous antimicro- bial polypeptides. J. Immunol. Methods 137, 167–173. 15. Johnson Jr, W.C. (1990) Protein secondary structure and circular dichroism: a practical guide. Proteins 7, 205–214.

Should a (cid:212)take-home message(cid:213) be needed from the present study, one might remember Gulliver and the Lilliputians who shackled him while he slept. Simply, that a multitude of small binding events, especially when they are performed cooperatively, can secure large objects.

16. Chen, Y.H., Yang, J.T. & Chau, K.H. (1974) Determination of the helix and beta form of proteins in aqueous solution by circular dichroism. Biochemistry 13, 3350–3359.

17. Guntert, P., Mumenthaler, C. & Wuthrich, K. (1997) Torsion angle dynamics for NMR structure calculation with the new program DYANA. J. Mol. Biol. 273, 283–298.

A C K N O W L E D G E M E N T S

18. Heiden, W., Moeckel, G. & Brickmann, J. (1943) A new approach to analysis and display of local lipophilicity/hydrophilicity map- ped on molecular surfaces. J. Comput. Aided Mol. Des. 7, 503–514. 19. Laskowski, R.A., Rullman, J.A., MacArthur, M.W., Kaptein, R. & Thornton, J.M. (2001) AQUA and PROCHECK-NMR: program for checking the quality of protein structures solved by NMR. J. Biomol. NMR 8, 477–486. This study was supported by grants from the National Institutes of Health: AI-43934 and AI-37945 to R. I. L., and HL-61234 to P. M. and B. T. The work was also supported, in part, by the Cystic Fibrosis Foundation (McCray, 97 ZO; Lehrer, 97-ZO) and by intramural funds from the University of Iowa Department of Medicine. We thank Reviewer 2 for suggesting the binding study with a mixture of truncated SMAP-29 peptides.

SMAP-29: conformation and LPS binding (Eur. J. Biochem. 269) 1189

(cid:211) FEBS 2002

outer membranes of gram-negative bacteria. Biochemistry 38, 13643–13653. 20. Buck, M. (1998) Trifluoroethanol and colleagues: cosolvents come of age. Recent studies with peptides and proteins. Q. Rev. Biophys. 31, 297–355.

29. Zhang, L., Dhillon, P., Yan, H., Farmer, S. & Hancock, R.E. (2000) Interactions of bacterial cationic peptide antibiotics with outer and cytoplasmic membranes of Pseudomonas aeruginosa. Antimicrob. Agents Chemother. 44, 3317–3321. 21. Goormaghtigh, E., Raussens, V. & Ruysschaert, J.M. (1999) Attenuated total reflection infrared spectroscopy of proteins and lipids in biological membranes. Biochim. Biophys. Acta 1422, 105–185. 22. Wuthrich, K. (1986) NMR of Proteins and Nucleic Acids. John Wiley & Sons, New York.

23. Din, Z.Z., Mukerjee, P., Kastowsky, M. & Takayama, K. (1993) Effect of pH on solubility and ionic state of lipopolysaccharide obtained from the deep rough mutant of Escherichia coli. Biochemistry 32, 4579–4586. 30. Moore, R.A. & Hancock, R.E. (1986) Involvement of outer membrane of Pseudomonas cepacia in aminoglycoside and polymyxin resistance. Antimicrob. Agents Chemother. 30, 923–926. 31. Hoess, A., Watson, S., Siber, G.R. & Liddington, R. (1993) Crystal structure of an endotoxin-neutralizing protein from the horseshoe crab, Limulus anti-LPS factor, at 1.5 A˚ resolution. EMBO J. 12, 3351–3356.

32. Beamer, L.J., Carroll, S.F. & Eisenberg, D. (1998) The BPI/LBP family of proteins: a structural analysis of conserved regions. Protein Sci. 7, 906–914. 24. Ferguson, A.D., Hofmann, E., Coulton, J.W., Diederichs, K. & Welte, W. (1998) Siderophore-mediated iron transport: crystal structure of FhuA with bound lipopolysaccharide. Science 282, 2215–2220.

33. Beamer, L.J., Carroll, S.F. & Eisenberg, D. (1997) Crystal struc- ture of human BPI and two bound phospholipids at 2.4 angstrom resolution. Science 276, 1861–1864. 25. Schindler, M. & Osborn, M.J. (1979) Interaction of divalent cations and polymyxin B with lipopolysaccharide. Biochemistry 18, 4425–4430. 26. Snyder, S., Kim, D. & McIntosh, T.J. 34. Tan, N.S., Ho, B. & Ding, J.L. (2000) High-a(cid:129)nity LPS binding domain(s) in recombinant factor C of a horseshoe crab neutralizes LPS-induced lethality. FASEB J. 14, 859–870.

(1999) Lipopoly- saccharide bilayer structure: effect of chemotype, core mutations, divalent cations, and temperature. Biochemistry 38, 10758– 10767.

35. Tan, N.S., Ng, M.L., Yau, Y.H., Chong, P.K., Ho, B. & Ding, J.L. (2000) Definition of endotoxin binding sites in horse- shoe crab factor C recombinant sushi proteins and neutralization of endotoxin by sushi peptides. FASEB J. 14, 1801–1813.

27. Chen, C., Brock, R., Luh, F., Chou, P.J., Larrick, J.W., Huang, R.F. & Huang, T.H. (1995) The solution structure of the active domain of CAP18 – a lipopolysaccharide binding protein from rabbit leukocytes. FEBS Lett. 370, 46–52.

28. Gutsmann, T., Larrick, J.W., Seydel, U. & Wiese, A. (1999) Molecular mechanisms of interaction of rabbit CAP18 with 36. Wishart, D.S., Sykes, B.D. & Richards, F.M. (1992) The chemical shift index: a fast and simple method for the assignment of protein secondary structure through NMR spectroscopy. Biochemistry 31, 1647–1651.