Tài liệu Báo cáo Y học: Antimicrobial dendrimeric peptides pot

Antimicrobial dendrimeric peptides James P. Tam, Yi-An Lu and Jin-Long Yang Vanderbilt University, Department of Microbiology and Immunology, MCN A5119, Nashville, TN, USA Dendrimeric peptides selective for microbial surfaces have been developed t o a chieve broad a ntimicrobial activity and low hemolytic activity to human erythrocytes. The dendri- meric core is an asymmetric lysine branching tethered with two to eight copies of a tetrapeptide (R4) or an octapeptide (R8). T he R4 tetrapeptide (RLYR) contains a putative microbial surface r ecognition B HHB motif ( B ¼ basic, H ¼ hydrophobic amino acid) found in protegrins a nd tachyplesins whereas the octapeptide R8 (RLYRKVYG) consists of an R4 and a degenerated R4 repeat. Antimicro- bial assays against 10 organisms in high- and low-salt con- ditions showed that the R4 and R8 monomers a s well as their divalent dendrimers contain no to low activity. In contrast, the tetra- and octavalent R 4 a nd R8 dendrimers are broadly active under either c onditions, exhibiting relatively similar potency with minimal inhibition concentrations < 1 l M against both bacteria and fungi. Based o n their size and charge similarities, the potency and activity spectrum of t he tetravalent R4 dendrimer are comparable to protegrins and tachyplesins, a family of potent a ntimicrobials containing 17–19 residues. Compared with a series of linearly repeating R4 peptides, the R4 dendrimers show comparable anti- microbial potency, but ar e m ore aqueous soluble, more stable to proteolysis, less toxic t o human cells and m ore easily synthesized chemically. These results suggest repeating peptides that cluster the charge and hydrophobic residues may represent a primitive form of microbial pattern-recog- nition. Incorporating s uch knowledge in a dendrimeric design therefore presents an attractive approach for devel- oping novel peptide antibiotics. Keywords: dendrimeric peptide; p eptide antibiotics. Cationic antimicrobial peptides constitute an important component of the innate immunity against microbial infections [1–6]. Recently there is renewed interest in developing novel approaches for designing peptide-based antibiotics m anifested by k illing mechanisms t hat are less likely than conventional antibiotics to develop multidrug resistance [7–12]. Design elements desirable for therapeutics include activity under physiological conditions (100– 150 m M or high-salt c on ditions), low t oxicity and proteo- lytic stability. Guided by these c onsiderations, we and others have designed antimicrobial peptides with unusual struc- tural a rchitectures using rigid sc affoldings such as cyclic peptides highly constrained with a cystine-knot motif on two or three b strands [10–12] to cluster hydrophobic and charge regions that produce amphipathic structures impor- tant for antimicrobial activity. Furthermore, these constraints confer metabolic stability, and impart mem- branolytic selectivity that minimizes toxicity. Another approach for designing antimicrobial peptides is based o n their mechanisms of action. An example would exploit mechanisms of recognizing conserved motifs on microbial surfaces that are not found in higher eukaryotes. Janeway & Medzhitov [13] have recently classified a family of proteins and receptors specific for pathogen associated molecular patterns (PAMPs) essential for innate and adaptive responses. P athogen-associated motifs include various microbial cell-wall components such as lipopoly- saccharide ( LPS), peptidioglycans, teichoic acids, mannans, N-formyl peptides, and lipidated peptides [14,15]. Some well-studied motif-recognizing proteins i nclude LPS -bind- ing p rotein, soluble and mem brane-anchored CD14 and Toll-like LPS receptors as well as mannose-binding protein and the receptors for mannans and manoproteins [16–18]. Cationic antimicrobial peptides may have also evolved t o recognize PAMPs on microbial s urfaces. T hey o ften possess a broad spectrum of a ntimicrobial activities against bacte- ria, fungi or viruses through mechanisms that generally involve the disruption of microbial envelopes. In general, at their effective killing doses, most antimicrobial peptides are nontoxic to host cells, s uggesting pattern-recognition selec- tivity under evolutionary pressure. Although more t han 200 antimicrobial peptides with various types of structures are known, they can b e classified into two broad categories based on their primary sequences: those that contain repeating sequences ranging from two to 14 amino a cids and those that are nonrepeating [19,20]. Found in these two types of peptides a re basic amino acids useful f or electro- Correspondence to J. P. Tam, Vanderbilt University, Department of Microbiology and Immunology, A-5119 MCN, 1161 21st A venue South, Nashville, TN 37232-2363, USA. Fax: + 1 615 343 1467, Tel.: + 1 615 343 1465, E-mail: james.tam@mcmail.vanderbilt.edu Abbreviations:CHCA,a-cyano-4-hydroxycinnamic acid; DCC, N,N-dicyclohexylcarbodiimide; DCM, dichloromethane; DIC, N,N-diisopropylcarbodiimide; DIEA, N,N-diisopropylethylamine; DMF, dimethylformamide; EC 50 , peptide concentration causing 50% hemolysis; Fmoc, 9-fluorenylmethyloxycarbonyl; Fmoc-DPA, p-(R,S)-a-[1-(9H-fluoren-9-yl)methoxyformamide]-2,4-dimethoxy- benzylphenoxyacetic acid; HOBt, N-hydroxybenzotriazole; LPS, lipoplysaccharide; MBHA resin, methylbenzhydrylamine resin; M IC, minimal inhibition concentration; PAMPs, pathogen associated molecular patterns; PG-1, protegrin-1; R t , retention time; RTD-1, rhesus theta defensin; TP-1, tachyplesin-1; TCEP, tris(carboxyethyl) phosphine; TSB, trypticase soy broth; SPPS, solid-phase peptide synthesis. (Received 2 October 2001, revised 2 December 2001, accepted 5 December 2001) Eur. J. Biochem. 269, 923–932 (2002) Ó FEBS 2002 static interactions with microbial membranes. Other amino acids useful for structural and hydrophobic roles have also been observed including Pro, Phe, and Trp [21,22]. A structural feature commonly found in antimicrobial pep- tides, whether t hey contain repeating or nonrepeating sequences, is their ability to cluster charge and hydrophobic amino a cids as amphipathic molecules to interact with the negatively charged lipidic microbial surfaces. We reasoned that this amphipathic structure might function partly for pattern recognition. Thus, we have explored an approach to exploit the polyvalency of a dendrimer to tether an array of short Ôpattern-recognition Õ peptides frequently found in b-stranded peptide antibiotics t o enhance interactions with microbial lipid surfaces. A s s hort peptides < 12 amino acids without conformational constraints are not likely to fold into a stable structure that provide strong antimicrobial actions under physiological conditions containing 100 m M NaCl, we also incorporated a dendrimeric design with a lipid-like backbone that facilitates interaction with microbial surfaces. Dendrimeric peptides or peptide dendrimers are biopoly- mers of unusual a rchitectures. First evolved in the 1980s [23,24], they contain a multivalent core that tethers an array of bran ching peptides. An early e xample i s the multiple antigen peptides (MAPs) introduced by our laboratory as immunogens for producing site-specific polyclonal and monoclonal antibodies. The dendrimeric core of a MAP (Fig. 1) consists of a d ivalent L ys core whose a and e amines double geometrically with each branching generation. Although a single lysine core with two amino moieties has been extensively used to design divalent-branched bioac tive peptides [25–27], a MAP-like dendrimeric peptide contain- ing the Lys 2 Lys (K 2 K) as a dendrimeric core is better suited to serve t he purpose o f providing with short peptides polyvalency for our dendrimeric design. Little is known a bout the structures of dendrimeric peptides with the K 2 K cores. The three-lysine K 2 Kcoreis asymmetric because each lysine c ontains a short a and a long e amino arm that results in unequally spaced amines with four branches, t wo e and two a amines. This combination of a and e peptide has an e-peptide backbone, which is torsionally flexible, with five methylenes separating the amine from the carboxylic acid. In an extended form, the en d-to-end d istance of 21 atoms separating the two e-amines of the K 2 K core, most of which are methylen e units mimicking the lipid chains, can be considered as lipid-like biopolymers sufficient for transversing a lipid membrane. Under a lipidic environment, modeling shows that a K 2 K core with tethered short a p eptides can achieve a barrel-like structure mimicking those o f helical peptides, which is essential for membranolytic activity. In addition, judicious selection of a short a peptide with cationic c harged and hydrophobic residues appropriate for Ôpattern recognitionÕ may enhance t he hypothetical bioactive structures through electrostatic and hydrophobic i nteractions with the n ega- tively charged microbial surfaces. Tachyplesins [28] are potent broad-spectrum antibiotics containing four degenerated repeats of a t etrapeptide with a HBCH motif (H, hydrophobic; B, basic a nd C, Cys). These cystine-stabilized antiparallel peptides have s ide chains arranged in an up-and-down topology. On e face o f this topology contains a consensus B HHB motif. Similar BHHB motifs can be found in protegrins [29] and RTD-1 (rhesus monkey theta defensin) [30]. Based on this topological motif, our prot otypic pattern-recognition peptides c onsist of a t etrapeptide R4, RLYR, and a degenerated double- repeating o ctapeptide R 8, RLYR-KVYG (Fig. 2). Here, we describe the d esign and properties of dendrimeric pep tides employing the tetravalent and octavalent dendrimeric cores tethered with cationic p eptides, a R4 tetrapeptide or a R8 octapeptide. For comparison, the R4 or R 8 peptides a re then tethe red t o three different cores consisting of a single lysine (K), a three-lysine (K 2 K) or a heptalysine [(K 2 K) 2 K] core to give divalent, tetravalent or octavalent dendrimeric peptides, respectively. The R4 dendrimeric peptides are also compared with a controlled series of linearly repeating R4 peptides of (RLYR) n , n ¼ 1, 2, 4 and 8. Our results show Fig. 1. Schematic representations of three type of dendrimeric cores with three generation of lysines shows in t hree different font style. (A) t wo branched Lys (bold ); (B) four branched (Lys) 2 Lys (regular); (C ) eight branched [(Lys) 2 Lys] 2 Lys (bold italic); (D) (Lys) 2 Lys core w ith a and e branche bearing peptides Arg-Leu-Tyr-Arg. Fig. 2. Topological distributions of BHH B and BHHX motif i n the antiparallel b sheet s tructured protegrin-1 (PG-1), tachyplesin-1 (TP-1) and rhesus theta defensin (RTD-1). 924 J. P. Tam et al. (Eur. J. Biochem. 269) Ó FEBS 2002 that the tetravalent and octavalent dendrimeric short peptides display p roperties desirable in antimicrobials. They are broadly active with very similar potency against 10 test organisms, but are also less toxic and more proteolytic resistant than the corresponding linearly repeating peptides. MATERIALS AND METHODS Materials Solvents, all of HPLC grade, were obtained from VWR Scientific C o. and used without further purification. Fmoc amino acid derivatives, N-hydroxybenzotriazole (HOBt), N,N¢-dicyclohexyl carbodiimide (DCC), and p-[(R,S)-a-[1-(9H-Fluoren-9-yl)-methoxylformamido]-2,4- dimethoxylbenzyl]-phenoxyacetic acid (Fmoc-DPA) were obtained from Chem-impex I nternational Inc. N,N-diiso- propylethylamine (DIEA) a nd p-cresol w ere purchased from Aldrich Chemical Co. Trypsin, a-chymotrypsin a nd a-cyano -4-hydroxycinnamic acid were purchased from Sigma Chemical C o. Trifluoroacetic acid was obtain ed from Halocarbon. Tris(carboxyethyl)-phosphine (TCEP) was obtained from Calbiochem. Ten organisms obtained from the American Type Culture Collection (ATCC, Rockville, MD, USA) were used for antimicrobial assays. Four Gram-negative bacteria included Escherichia c oli ATCC 25922, Pseudomon as aeru- ginosa ATCC 27853, Klebsiella oxytoca ATCC 49131, and Proteus vulgaris ATCC 49132. The three Gram-positive bacteria were Staphylococcus aureus 29213, Micrococcus luteus ATCC 49732 and Enterococcus faecalis ATCC 29212. The three fungi were Candida albicans ATCC 370 92, Candida kefyr ATCC 37095, and Candida tropicalis ATCC 37097. The s trains were incubated in trypticase soy broth (TSB) which was p repared in double d istilled water and autoclaved for sterilization. TSB was purchased from Becton–Dickinson (Cockeysville, MD, USA). Peptide syntheses and purification Solid-phase peptide synthesis on an automated s ynthesizer (CS Bio Co. 536) was performed using Fmoc-tBu chemistry and a sin gle coupling protocol with B OP/DIEA [31] in DMF. Linear peptides were synthesized on F moc-DPA- resin a nd dendrimeric peptides wer e synthesized on d ifferent branching core matrix Fmoc-DPA-NH-resins. A nalytical RP-HPLC was conducted on a Shimadzu L G-6 A system with an C 18 Vydac column ( 4.6 · 250 mm). A linear gradient of 10–90% buffer B ran for 30 min at 1 mLÆmin )1 with detection at 225 nm. Eluen t A contained 0.04% TFA/ H 2 O; eluant B contained 0.04% trifluoroacetic acid/60% CH 3 CN/H 2 O. Preparative RP-HPLC was performe d on a Waters 600 system with C 18 Vydac column ( 20 · 250 mm). MALDI-MS was measured on a Pe rS eptive Biosystems Voyager instrument. Samples were d issolved in 1 lLofa 1:2mixtureofH 2 O/CH 3 CN. Measurements were m ade in a linear m odel, with a-cyano-4-hydroxycinnamic acid a s the matrix. Preparation of Fmoc-DPA-NH-resin [32] The amine resin (2 g, 0 .22 mmol) was first swollen and washed with DCM. Fmoc-DPA (539 mg, 1 mmol) in DMF (40 mL) was added, shaken for 5 min ( not drained), a nd followed by D CC (206 mg, 1 mmol) and HOBt (157 mg, 1 mmol) to anchor the F moc-DPA handle onto the resin support. After 24 h, the resultant Fmoc-DPA-resin was drained, was hed sequen tially with DMF, DCM, MeOH and dried in vacuo. Substitution of the functionalized resin was 0.1 mmol Æg )1 (Fig. 3). Preparation of branching lysyl core resin Syntheses of d i- tetr a- and octa-branching cores required one, t wo and three coupling cycles, respectively, using a four molar excess of Fmoc-Lys (Fmoc) via BOP/DIEA in DMF on the Fmoc-DPA-NH-resin. The Fmoc group was removed b y treatment with 20% piperidine/DMF. Each coupling cycle doubled the branching level of lysyl core and after three cycles, the octa-branching Fmoc-Lys 4 -Lys 2 -Lys- DPA-NH-resin was achieved (Fig. 3). Synthesis of DmR4 and DmR8 These peptides were synthesized on different branching lysyl c ore resin according to what has been described above. Th e HPLC retention time and m ass data showed in Table 1. Fig. 3. Synthetic scheme for preparing different branched (two, four, and eight) core. Lo w substituted a mine resin 1 coupled with DP A 2 by DCC/HOBt in DMF to f ormFmoc -DPA- resin 3. After deprotection, 3 was coupled to Fmoc-Lys(Fmoc)-OH throughBop/DIEA to form two branching core resin 4. Repeating one or two rounds of c ouplings with Fmoc- Lys(Fmoc)-OH fo rmed f our b ranchin g c ore 5 or eight branching core 6. Ó FEBS 2002 Antimicrobial dendrimeric peptides (Eur. J. Biochem. 269) 925 Antimicrobial and hemolytic assays A sensitive and reproducible two-stage radial d iffusion antimicrobial assay of Lehrer et al .[33]wasemployed. Antimicrobial activities were expressed in units (0.1 mm ¼ 1 U ), and the MICs were determ ined from the x-intercepts of the dose–response curves. Hemolytic activity was determined u sing fresh human er ythrocytes. Peptide concentrations causing 50% hemolysis (EC 50 )were derived from the dose–response curve [34]. The membrano- lytic selectivity index is expressed as EC 50 /MIC. Proteolytic stability Dendrimeric o r linear peptides i n various concentrations were dissolved in NaCl/P i buffer at pH 7.4 and aliquoted into microtubes. Trypsin was m ixed with peptides in the ratio of 1 : 100 (enzyme/peptide, w/w). The enzymatic degradation was carried out at 37 °C and stopped b y adding an appropriate enzyme inhibitor such as Typ e II-S trysin inhibitor i nto the samples. The r esidual a ntimicro- bial activity of each sample was determined by a two - stage r adial diffusion assay using E. coli as previously described [33]. The d iameter o f t he clear z one of control (nonenzyme treatment) is d efined as 100% active a nd the antimicrobial activity of samples is expressed as percent- age o f c on trol. RESULTS Design and synthesis The tetrapeptide R4 and the o ctapeptide R8 were derived from the topological motifs of the cystine-stabilized b-stranded a ntimicrobial peptid es, PG ( protegrins), TP (tachyplesins) and RTD-1 (rhesus monkey t heta defensin). These n aturally occurring peptides consisting of 17–19 residues containing two antiparallel b stran ds stabilized by two or three cross-strand disulfides that rigidify an up-and- down side-chain arrangement with to p a nd bottom faces [35–37]. The arbitrarily assigned bottom hydrophobic faces of protegrins, tachyplesins and RTD-1 are packed with two or three disulfide bonds, while a spatially arranged BHHB motif ( B ¼ basic a nd H ¼ hydrophobic amino acids) can be found on their top faces. These include tetrapeptides KWFR and RVYR in tachyplesin-1, RLYR in protegrin-1 and RITR in RTD-1 (Fig. 2 ). For convenience, we selected RLYR of protegrin-1 a s the consensus BHHB sequence for designing the R4 tetrapeptide. The octapeptide R 8, Arg- Leu-Tyr-Arg-Lys-Val-Tyr-Gly, was designed to contain a degenerated BHHB double-repeat with alternating clusters of charge regions and hydrophobic a mino acids. The double charge motif of Arg-Arg is found spatially or contiguously in TP-1, PG-1 and RTD-1 w hile the degenerate tetrapeptide repeat (KVYG) containing a single b ase and three hydro- phobic amino acid sequences are also f ound as retro- sequences in TP-1 (RYIG), and in PG-1 (RVVF). Two dendrimeric and a linearly repeating peptide series ranging f rom four to 32 amino acids were prepared (Table 1). The dendrimeric series contained R 4 and R8 peptides tethered on an asymmetric core as dendrimeric peptide D m R4 and D m R8 with branch numbers (m) of t wo, four, and eight, respectively. Two dendrimeric peptide series were compared with a series c ontaining linearly repeating R4 peptides (R4) n with n ¼ 1, 2, 4 and 8. All p eptides were prepared chemically by a stepwise solid- phase method [38] purified by RP-HPLC a nd characterized by mass spectrometry. An advantage of a dendrimeric over a linear peptide of comparable molecular size is that they require far fewer steps in their chemical synthesis. As peptide dendrimers a re prepared by a controlled polymer- ization in w hich multip le peptides copes grow simultane- ously on the branching cores, the number of assembling steps in their syntheses is sharply reduced when comparing to linear p eptides o f similar lengths (Fig. 2). Thus, the R8 dendrimeric peptides required only eight cycles (120 steps) by a solid-phase method for assembly on the lysyl cores to afford th ree D m R8 dendrimers containing 16, 32 a nd 64 amino-acid residues, respectively. Three dendrimeric R4 peptides required only four cycles. In contrast, the similarly sized 32-residue (R4) 8 peptide of the linearly repeating R4 series required 3 2 c ycles and 480 steps for its synthesis, which clearly indicates th e synthetic advantage in preparing the dendrimeric series. Table 1. S equences, HPLC retention time, mass data and number of monomer, a mino acid and charge of linear and dendrimeric peptides. Compounds Sequence a HPLC (min) MH + (calculated) Monomer Number of Amino acids b Charge c A Linear R4 1 R4 RLYR 11.3 605.518 (607.691) 1 4 2 2 (R4) 2 (RLYR) 2 12.8 1193.83 (1194.42) 2 8 4 3 (R4) 4 (RLYR) 4 17.1 2369.66 (2371.81) 4 16 8 4 (R4) 8 (RLYR) 8 23.9 4726.71 (4726.84) 8 32 16 B Dendrimer R4 5D 2 R4 (RLYR) 2 -[K] 13.2 1322.44 (1322.55) 2 8 4 6D 4 R4 (RLYR) 4 -[K 2 K] 14.5 2756.01 (2756.27) 4 16 8 7D 8 R4 (RLYR) 8 -([K 2 K] 2 K) 17.2 5623.44 (5623.71) 8 32 16 C Dendrimer R8 8 R8 RLYRKVYGK 11.8 1182.72 (1181.43) 1 8 3 9D 2 R8 (RLYRKVYG) 2 -[K] 14.8 2220.23 (2218.52) 2 16 6 10 D 4 R8 (RLYRKVYG) 2 -[K 2 K] 15.7 4549.02 (4549.55) 4 32 12 11 D 8 R8 (RLYRKVYG) 2 -([K 2 K] 2 K) 17.3 9214.54 (9211.27) 8 64 24 a C-terminal amide. b Excluding the lysyl core. c Charge refers to the dendrimeric peptides and the lysyl core. 926 J. P. Tam et al. (Eur. J. Biochem. 269) Ó FEBS 2002 Antimicrobial activity The minimum inhibitory concentrations (MICs) of all three series against four Gra m-negative bacteria, three G ram- positive bacteria and three fungi were determined by two- stage radial diffusion assay in both low- and high-salt (with 100 m M NaCl) conditions. Assays of antimicrobial activity under high-salt conditions were intended to simulate physiological conditions of 100–150 m M NaCl. Dendrimeric RLYRKVYG (R8) octapeptides Under low-salt conditions, the R8 monomer containing an additional charged Lys at its C-terminus showed low activity against Gram-positive and Gram-negative b acteria as well as m oderate activity against t hree tes t fungi w ith MICs > 10 l M (Table 2). However, their activities against all t est organisms were completely abrogated under high- salt condition s. The potency of the dimeric 16-r esidue peptide D 2 R8 was significantly i mproved, with MICs <10 l M against seven organisms and < 1 l M against E. coli and M. luteus under low-salt condition. D 2 R8 did not retain its activity under high-salt conditions, except against M. luteus (MIC, 1.8 l M ). In contrast, the tetravalent D 4 R8 was broadly active under both l ow- and high-salt conditions. With the exception of D 4 R8 against Pr. vulgaris (MIC, 2 l M ),theirMICswere<1l M under low-salt conditions. T he tetr avalent D 4 R8, w ith a mean MIC o f  0.67 l M against all test organisms under both low- and high-salt conditions, displayed activity spectra and potency comparable to naturally occurring antimicrobial peptides such as tachyplesins and protegrins. Dendrimeric RLYR (R4) tetrapeptides Comparing with the previous R8 series, the dendrimeric R4 series h as t he adv antag es o f m olecular and synthetic simplicity. Because the R4 peptides contain only the first repeat of the R8 octapeptides, their molecular sizes and the chemical steps needed for their syntheses are quantitatively halved. However the d endrimeric R4 peptides of eight, 16 and 32 amino acids with valences of two, four and eight are nearly as potent as the corresponding D m R8 series. The R 4 monomer was inactive with MICs > 500 l M against 10 test organisms in both low- and high-salt assays (Table 3). In contrast, the eight-residue divalent D 2 R4 displayed c onsiderable activity with MICs ranging from 1 to 6.1 l M and a mean MIC of 3.3 l M under low-salt conditions, averaging > 160-fold improvement over the R4 tetrapeptide. However, its activity decreased twofold to 10-fold under high-salt conditions. Interestingly, D 2 R4 was selective against three test fungi with MICs of 1–1.3 l M under low-salt conditions. T etra-branching to a 16-residue D 4 R4 dendrimeric peptide increased potency  fivefold with MICs 0.39–1 l M under low salt-conditions and 0.59–1.9 l M under high-salt conditions. Based on the similarity of lengths, this 16-residue D 4 R4 is sixfold and 40-fold more active than the corresponding divalent 16-residue D 2 R8 of the R8 s eries a nd is comparable to PG-1 under low- and high-salt conditions, respectively. The D 8 R4 displayed MICs < 0.79 l M against all 10 test organisms under low-salt conditions and MICs < 1 l M against eight organisms under high-salt conditions. Further branching to t he octavalent 32-residue D 8 R4ledtoonly small improvements in potency, n early all which was retained under high-salt conditions. Linearly repeating RLYR (R4) tetrapeptides A series o f linearly repeating R4 peptides containing two, four, and eight copies of R4 peptides also was prepared for comparisons with the D m R4 dendrimers. In general, the antimicrobial a ctivity of the linear p eptides improved as molecular size increased. The linear octapeptide (R4) 2 exhibited M ICs ranging widely from 1.2 to 3 9 l M higher activity against Gram-positive t han Gram-negative bacte- ria, but was less active than the corresponding dendrimeric D 2 R4 peptide (Table 4). The tetrameric (R4) 4 with four R4 repeats a nd 16 residues showed MICs of 0.5–1.8 l M , but was also less active than t he corresponding D 4 R4 and D 8 R4 Table 2. Antimicrobial activity of dendritic R8 peptides. Experiments w ere pe rformed in radial diffusion assay with underlay gel containing 1% agarose, 10 m M phosphate buffer with (high-salt) or without ( low-salt) 100 m M NaCl. Activities against multiple strains are expressed as t he minimum inhibitory concentration (MIC, l M ). Organism MIC (l M ) R8 D 2 (R8) D 4 (R8) D 8 (R8) LH LHLHLH Gram-negative E. coli 28.4 > 500 0.8 29.8 0.2 2.0 0.3 0.3 P. aeruginosa 46.2 > 500 6.3 40.4 0.6 0.6 0.5 0.5 P. vulgaris 32.0 > 500 18.4 117 2.0 5.0 0.7 0.5 K. oxytoca 99.8 > 500 1.4 29.6 0.5 2.0 0.4 0.5 Gram-positive S. aureus 72.4 > 500 2.1 118 0.7 2.1 0.3 0.4 M. luteus 28.4 > 500 0.8 1.8 0.2 0.5 0.2 0.4 E. faecalis 10.3 > 500 5.0 17.2 0.7 0.8 0.3 0.4 Fungi C. albicans 19.2 > 500 1.9 14.4 0.6 1.7 0.3 0.4 C. kefyr 16.2 > 500 3.1 38.2 0.7 1.0 0.4 0.4 C. tropicalis 10.1 > 500 5.0 28.8 0.5 0.5 0.4 0.4 Ó FEBS 2002 Antimicrobial dendrimeric peptides (Eur. J. Biochem. 269) 927 dendrimers. Comparing with the D m R4 series, linear R4 peptides were generally less act ive against many tested organisms. The activity profile of octameric (R8) 8 could not be determined accurately because it precipitated in phos- phate buffers, suggesting aggregation under both low- and high-salt conditions. Hemolytic activity Table 5 shows the hemolytic activities of t he dendrimeric or linear R4 a nd R8 pep tide s eries expressing E C 50 values ranging > 500-fold. The toxicity of linear peptides against erythrocytes was g enerally higher than the dendrimeric R4 peptides. The monomeric R4 and R8, peptides which were ineffective as a ntimicrobials, were also nonhemolytic with EC 50 >3900l M . The hemolytic toxicity of the linear R4 peptides increased 2 .8- and 15-fold f rom R4 to t he dimer and tetramer, respectively. In contrast, the hemolytic activity of the dendrimeric R4 peptides increased only 2.3-and3.3-foldfromR4toD 2 R4 and D 4 R4 peptides, respectively. Int erestingly, the hemolytic activit y of the dendrimeric D 8 R4 (EC 50 1514 l M ) was similar to D 4 R4 (1510 l M ). Based on the molecular size, the t oxicity o f ( R4) 2 peptide on human erythrocytes was t wofold higher than D 2 R4, w hile the ( R4) 4 peptide was about fourfold higher than the corresponding D 4 R4. A significant d ifference in the effects on e rythrocytes m orphology o f was als o observed between the linear and dendrimeric peptides when the peptides and cells were incubated at 37 °C. The higher ordered linear R8 peptides (R4) 4 and (R4) 8 caused cell rufflings and aggregations which were n ot found with in the corresponding dendrimeric R8 peptides. Although t he EC 50 of the linear peptide (R 4) 8 could n ot be determined accurately because o f its poor solubility in phosphate buffers, this peptide rapidly and quantitatively induced erythrocytes aggregations. Proteolytic stability The proteolytic stability of d endrimeric peptides to trypsin and chymotrypsin was determined using peptide and enzyme in ratio of 100 : 1 (w/w) at 3 7 °C. Enzyme-treated Table 4. Antimicrobial activity of linear R4 peptides. Experiment were performed in radial diffusion assay as described for Table 2. Organism MIC (l M ) (R4) 2 (R4) 4 L-salt H-salt L-salt H-salt Gram-negative E. coli 39.0 33.6 1.0 0.7 P. aeruginosa 4.1 8.7 0.5 1.0 P. vulgaris 10.2 18.4 1.8 1.2 K. oxytoca 14.1 13.1 0.9 0.9 Gram-positive S. aureus 8.1 17.5 1.8 1.1 M. luteus 1.3 12.8 0.6 0.6 E. faecalis 1.2 5.2 0.9 1.4 Fungi C. albicans 9.0 17.4 1.2 1.9 C. kefyr 1.7 5.5 1.1 2.0 C. tropicalis 1.4 3.2 0.8 1.2 Table 3. Antimicrobial activity of R4 dendrimer peptides. Experiment were performed in radial d iffusion assay as described for Table 2. MIC (l M ) R 4 D 2 (R4) D 4 (R4) D 8 (R4) Organism L- salt H-salt L- salt H-salt L- salt H-salt L- salt H-salt Gram-negative E. coli > 500 > 500 6.1 10.2 0.6 0.7 0.5 0.7 P. aeruginosa > 500 > 500 3.6 18.5 0.5 1.2 0.3 0.9 P. vulgaris > 500 > 500 3.3 28.0 1.0 1.9 0.8 1.3 K. oxytoca > 500 > 500 3.8 39.0 0.4 0.9 0.4 0.8 Gram-positive S. aureus > 500 > 500 4.5 10.2 0.8 0.6 0.5 0.6 M. luteus > 500 > 500 3.9 12.0 0.5 0.7 0.5 0.7 E. faecalis > 500 > 500 4.7 10.2 0.8 1.8 0.8 1.3 Fungi C. albicans > 500 > 500 1.0 6.4 0.8 0.8 0.7 0.9 C. kefyr > 500 > 500 1.2 1.5 0.9 1.3 0.7 0.6 C. tropicalis > 500 > 500 1.3 1.8 0.7 0.8 0.7 0.6 Table 5 . H emo lytic activity of R4 and R8 linear and dendritic p eptid es. Hemolytic activity of peptides is expressed as EC 50 ,whichisthepep- tide concentration producing 50% of human erythrocytes lysis. Peptide EC 50 (l M ) Linear RLYR 5200 (RLYR) 2 1950 (RLYR) 4 338 (RLYR) 8 48 Dendritic R4 (RLYR) 2 K 3700 (RLYR) 4 K 2 K 1510 (RLYR) 8 (K 2 K) 2 K 1514 Dendritic R8 RLYRKVYG 3950 (RLYRKVYG) 2 K 1420 (RLYRKVYG) 4 K 2 K 610 (RLYRKVYG) 8 (K 2 K) 2 K 112 Tachyplesin 108 928 J. P. Tam et al. (Eur. J. Biochem. 269) Ó FEBS 2002 samples collected at different time p oints were tested for their residual antimicrobial activity against indicator bacterium E. coli (Fig. 4). When the linear (R4) 4 peptide was treated with trypsin, its antibacterial activity decreased rapidly to 25–30% of the untreated control after 2–5 min and remained at about 20% of control after 24 h. In contrast, t he ac tivity of the trypsin-treated D 4 R4 peptide dendrimer was > 9 0% and 80% of control after 2 and 24 h, respectively. Similar results were observed for the dendrimeric R8 peptides (Fig. 5). Trypsin r apidly inactivated > 80% of the antibacterial activity of the linear R8 peptides, but only  30% of D 2 R8 and D 4 R8 activity after 2 h. To determine whether these results derived from a loss of trypsin activity during t he assay, D 2 R8 was retreated with trypsin a t 4 - a nd 8- h i ntervals a fter the first trypsin treatment. Results showed that eac h treatment d ecrease d the antibacterial activity of D 2 R8 about 10–20%, while the antimicrobial activity of D 4 R8 decreased 1 5% after 8 h trypsin re-treatment. These findings suggest that the dendrimeric peptides were more resistant to proteolytic degradation. DISCUSSION We have demonstrated that dendrimers containing repeat- ing short tetrapeptides with a BHHB motif modified f rom naturally occ urring b-stranded antimicrobial peptides func- tion as potent antimicrobials with membranolytic activity. Although dendrimeric and linearly repeating peptides differ in their architectures and t opologies, they may share a similar ability to form various patterns of hydrophobic and charge clusters for pattern recognition of microbial surfaces. Compared with the linearly repeating peptide antimicro- bials, our results s how that the dendrimeric peptides possess several desirable attributes. They display potent and broad- spectrum activity under both low- and high-salt conditions, enhanced proteolytic stability and dec reased hemolytic activity. Furthe rmore, they require far fewer chemical steps for their synthesis than the control s eries of tandemly repeating peptides. Although the a ctivity of a ntimicrobial peptides such as defensins or defensin-like p eptides with M r ranging from 3 to 5 kDa is abrogated when tested in high-salt conditions, this is not true in peptide like p rotegrins, and tachyplesins with an M r of about 2 k Da [1–4]. Thus, these assays would show whether the dendrimeric peptides ranging from 2 to 5 kDa behave similar t o defensins or protegrins and tachyplesins and whether molecular s izes have any e ffect on their activity profiles. F or comparative purposes, Table 6 shows the antimicrobial activity of PG-1 and TP-1 in our assay system. In general, TP-1 with MICs ranging from 0.2 to 1.3 l M is the m ore potent of these two antimicrobial peptides, d isplaying twofold to threefo ld higher activity than PG-1 (MICs 0.3–2.8 l M )innineofthe10tested organisms in our assays. I t is also interesting to note that PG-1 displays higher activity against Gram-positive than Gram-negative bacteria or fungi under both low- and high- salt conditions with MICs ranging from 0.3 to 0.8 l M . Correlation of dendrimeric design with antimicrobial activity In the current study, th e optimal branching related to antimicrobial activity and molecular size suitable f or further development as therapeutics appears to be tetravalent. The R4 and R8 monomers in t he D m R4 and D m R8 series are largely inactive w hereas the dimeric forms do not retain their Fig. 4. Activity aga inst E. coli of metabolite residures of li near and dendrimeric R4 by trypsin treatment. The peptides were t reated with trypsin at 3 7 °C. At devising tim es, the s amples were collected and trypsin inhibitor was added to samples for stopping the reaction. The antimicrobial activity of each sample against E. coli was performed in a two-stage radial diffusion assay. The antimic robial activity of sam- ples is expressed i n percentage of that of samples w ithout trypsin treatment. Fig. 5. Ac tivity against E. coli of metabolite residures of dendrimeric R8 by trypsin treatment. Effect of trypsin on the anitmicrobial activity of D 2 R8, D 4 R8 and D 8 R8 peptides . Th e ex perimen ts w ere p erformed a s described in the legend of Fig. 4. Ó FEBS 2002 Antimicrobial dendrimeric peptides (Eur. J. Biochem. 269) 929 activities under high-salt conditions. In contrast, the tetra- valent and octavalent (m ¼ four and eight, respectively) dendrimeric peptides show a broad activity spectrum against 10 test microbes in both low- and high-salt assays. For both R4 and R8 peptides, tetravalent dendrimers show large improvements in potency over divalent dendrimers whereas only small improvements are found from the tetravalent to octavalent dendrimeric peptides under low-salt conditions. The higher branching D 8 -dendrimers a nd longer peptide chain lengths of R8 peptides in retain activity under high- salt conditions better than the corresponding D 4 -dendrimer and shorter R4 peptide series. As there are only small variations in potency (MICs < 1 l M ) between the tetrava- lent and octavalent dendrimers, a tetravalent dendrimeric D 4 R4 design is perhaps more promising for further research on antimicrobials using other tetrapeptide analogues. The corresponding controlled series of linear R4 p eptides exhibits large variations of activity sp ectra and potency that roughly c orrelate with the decreases i n their lengths. The linear R4 peptides w ith less than t hree repeats are largely inactive under high-salt conditions, except against E. coli. The potency and activity spectra of the D 4 R4 dendrimer containing four R4 tetrapeptide copies are comparable to the p rotegrins and tachyplesins of similar lengths. More significantly, the dendrimeric R4 peptide achieves a comparable antimicrobial profile without the conformation constraints found in tachyplesins and protegrins, whose activities are significant r educed in their unconstrained forms. Taken together, these results suggest that a dendri- meric scaffold c ould serve as a template for further analog studies using a combinatorial approach with short peptides to improve potency and specificity. Our previous attempts to exploit t he dendrimeric design on 33-residue a helical antimicrobial peptide cecropins to increase potency were disappointing. Tetrameric and octameric cecropins did not result in e nhanced potency or specificity. A plausible explanation is that cecropins form ordered ahelical structures in aqueous environments [39]. Detailed structural information would thus be necessary to determine their approximate quaternary structures by a dendrimeric design. Thus, dendrimeric antimicrobial peptides based o n s hort p eptides consisting of four amino acids may h ave to overcome this type of limitation. Hemolytic activity and proteolytic stability The dendrimeric R4 and R8 peptides are essentially nontoxic to human erythrocytes with EC 50 for h emolysis ranging from 112 to 3700 l M . For example, using the mean MICs of D 4 R4 and D 4 R8 of 0.7 l M and 0.6 l M , respect- ively, for all 10 test organisms, the t herapeutic indices (EC 50 / MIC) of the two dendrimers will be > 2200. These two dendrimeric peptides show about 10-fold improvement over the line ar ( R4) 4 peptide, which has an EC 50 of 338 l M and a therapeutic i ndex of 200. The EC 50 of peptide (R4) 8 cannot be determined because o f its poor solubility in phosphate buffers. Peptide (R4) 4 also causes aggregation and mem- brane r ufflings of erythrocytes at concentrations < 10 l M , which are not observed with the D 4 R4 and D 8 R4 at concentrations > 1500 l M . Together, these results suggest that the D m R4 or D m R8 peptides are nontoxic to human erythrocytes at their effective microbe-killing concentra- tions. However, t he mechanisms that cause t hese differences are not clear. D 4 R4 is surprisingly more stable than the (R4) 4 peptide to proteolysis. The antimicrobial activity of the A rg-rich (R4) 4 is inactivated by trypsin within 10 min while the D 4 R4 peptide retains > 80% of its activity after 24 h. These results suggest that either the dendrimeric structure of D 4 R4 is more resistant to proteolytic degradation or that it is a protease inhibitor. Because incubation of D 4 R4 with trypsin does not inhibit t rypsin activity against the degra- dation of a chromogenic trypsin substrate, it is likely that the proteolytic stability o f D 4 R4 is due to its dendrimeric structure. Advantages and potential applications of dendrimeric short peptides Low m olecu lar mass p ep tide dendrimers have the a dvan- tage of being less immunogenic than high-molecular-mass dendrimers. The multivalency of peptide dendrimers appears to be d esirable in the d esign of m embranolytic peptides for other biochemical applications. These include their a bility t o amplify cationic charges and hydrophobic clusters as the number o f d endrimer branches increases. Polycationic peptides, whether linear o r branched , are known to d isplay membrane disruption o r f usion properties that have been exploited for intracellular peptide, protein and gene delivery [ 40–44]. Hydrophobic clusters o n a peptide dendrimer lead to aggregation that may enhance fusogenic a ctivity. A plausible mechanism is t hat amplifica- tion by a dendrimeric design increases the effective molarity of monomeric units and decreases the entrop y of self- assembly. The end results are t hat t hey may mimic t he mechanisms of action through w hich high-ordered anti- microbial peptides exert their membranolytic effects. Thus, the short dendrimeric peptides may represent a u seful a nd unusual biopolymer design for effecting various membrano- lytic activities in lipid environments. Table 6. Antimicrobial activity of protegrin a nd tachyplesin. Experi- ment we re performed i n radial diffusion assay as described at Table 2. Organism MIC (l M ) Protegrin Tachyplesin L-salt H-salt L-salt H-salt Gram-negative E. Coli 0.9 0.8 0.3 0.4 P. aeruginosa 1.2 2.0 0.9 0.5 P. vulgaris 2.4 2.8 0.7 1.0 K. oxytoca 0.7 0.9 0.2 0.5 Gram-positive S. aureus 0.7 0.6 0.4 0.5 M. luteus 0.3 0.8 1.0 1.1 E. faecalis 0.3 0.7 0.3 0.4 Fungi C. albicans 1.3 1.0 0.7 0.9 C. kefyr 1.8 1.4 0.9 1.3 C. tropicalis 1.0 1.5 0.5 1.0 930 J. P. Tam et al. (Eur. J. Biochem. 269) Ó FEBS 2002 ACKNOWLEDGEMENT This work was supported, in p art, by US Public He alth Se rvice NIH Grants CA36544 and AI46164. REFERENCES 1. Boman, H.G. (1995) Peptide antibiotics and their role in innate immunity. Annu.Rev.Immunol.13, 61–92. 2. Nicolas, P. & Mor, A. (1995) Peptide as weapons against micro- organisms in the chemical d efense system of vertebrates. Annu. Rev. Microbiol. 49, 277–304. 3. Meister, M., L emaitre, B. & Hoffmann, J.A. (1997) Antimicrobial peptide defence in Drosophila. Bioassay 19, 1019–1026. 4. Hancock, R.E. (1997) Peptide antibiotics. Lancet. 34 9, 418–422. 5. Ouellette, A.J. & S elsted, M.E. (1996) Paneth cell defensins: endogenous peptide c ompone nts of intes tinal host defen ce. FASEB J. 10, 1280–1289. 6. Lehrer, R.I. & G anz, T. (1999) Antimicrobial peptide in mammalian a nd insect host defence. Curr. Opin. I mmunol. 11, 23–27. 7. S olomon, S., Hu, J., Zhu, Q., Belcourt, D., Bennett, H.P., Bateman, A. & Antakly, T . (1991) Corticostatic pe ptides. J. Steroid B iochem. Mol. B iol. 40, 391–398. 8. Chertov, O., Michiel, D.F., Xu, L., Wang, J.M., Tani, K., Murphy, W.J., Longo, D.L., Taub, D.D. & Oppe nheim, J. (1996) Identification of Defensin-1, Defensin-2, and CAP37/ Azurocidin as T-cell ch emo attractant pr oteins released f rom interleukin-8-stimulated neutrophils. J. Biol. Chem. 271 , 2935– 2940. 9. Xu, Y ., Tamamura, H., Arakaki, R., N akashima, H., Zhang, X., Fujii, N., Uchiyama, T. & Hattori, T. (1999) Marked increase in anti-HIV activity, as well as inhibitory activity against HIV entry mediated by CXCR4, linked to enhancement of the binding ability of tachyplesin analogs to CXCR4. AIDS Res. Human Retroviruses 15, 419–427. 10. T am, J.P., Lu, Y A. & Yang, J L. (2000) Design of salt-insensi- tive glycine-rich antimicrobial peptides with cyclic tricysti ne structures. Biochemistry 39, 7159–7169. 11. T am, J.P., Wu, C. & Yang, J L. (2000) Membranolytic selectivity of cystine-stablizedcyclic protegrins. Eur. J. Biochem. 267, 3289– 3300. 12. T am, J.P., Lu, Y A., Y ang, J L. & Chiu, K W. (1999) A n unusual structural motif of antimicrobial peptides containing end- to-end macrocycle and cystine-knot d isulfides. Proc. Natl Acad. Sci. USA 96, 8913–8918. 13. Medzhitov, R. & Janeway Jr, C. (2000) The toll receptor family and microbial recognition. Trends Microbiol. 8, 452–456. 14. A derem, A. & Ulevitch, R.J. (2000) Toll-like r ecept ors in the induction of the innate immune response. Nature 406, 782–787. 15. Muhle, S.A. & Tam, J.P. (2001) Design of Gram-negative selective antimicrobial peptides. Biochemist ry 40, 5777–5785. 16. S chumann, R.R., Leong, S.R., Flaggs, G.W., Gray, P.W., Samuel, D.,Wright,S.D.,Mathison,J.C.,Tobias,P.S.&Ulevitch,R.J. (1990) Structure and function of lipopolysaccharide bind ing pro- tein. Science 249, 1429–1431. 17. W right, S.D., Ramos, R.A., Tobias, P.S., Ulevitch, R.J. & Mathison, J.C. ( 1990) CD4, a receptor for complexed o f lipopolysaccharide (LPS) and L PS binding protein. Science 249, 1431–1433. 18. W illiams, M.J., Rodriguez, A., Kimbrell, D.A. & Eldon, E.D. (1997) The 18-wheeler mutation r eveals complex a ntibacterial gene regulation in Drosophila host defence. EMBO J. 16, 6 120– 6130. 19. Frank, R.W., Gennaro, R., Schneider, K. & Prz ybylski, M. (1990) Amino acid sequences of two proline-rich bactenecins. J. Biol. Chem. 265, 18871–18874. 20. S hamova, O., Brogden, K .A., Z hao, C., Nguyen, T., Kokryakov, V.N. & L ehrer, R. (1990) I P urification and p roperties o f p roline- rich antimicrobial peptides from sheep and goat leukocytes. Inf ect. Immun. 67, 4106–4111. 21. Zanetti,M.,Sal,G.D.,Storici,P.,Schneider,C.&Romeo,D. (1993) The cDNA of the n eutrophil antibiotic Bac5 predicts a pro-sequence homologous to a cysteine proteinase inhibitor that i s common to other neutrophil a ntibio tics. J. Biol. Chem. 268 , 522–526. 22. S elsted, M.E., Novotny, M.J., Morris, W.L., Tang, Y Q., Smith, W. & Cullor, J.S. (1992) Indolici din, a novel bactericidal tridecapeptide amide from neutrophils. J. Biol. Chem. 267, 4292– 4295. 23. T am, J.P. (1988) Synthetic peptide vaccine design: synthesis and properties of a h igh-density multiple a ntige nic peptide s ystem. Proc. Natl Acad. Sci. USA 85, 5409–5403. 24. Posnett, D.N., McGrath, H. & Tam, J.P. (1988) A novel method for producing anti-peptide antibodies: production of site- specific antibodies to the T cell antigen receptor b-chain. J. Biol. Chem. 263, 1719–1725. 25. F rancis, M.J., Hastings, G.Z., Brown, F. , McDermed, J., Lu, Y A. & Tam, J.P. (1991) I mmunological evaluation of the multiple antigen peptide (MAP) s ystem using the major immu- nogenic site o f f oot-and mouth d isease v irus. Immunology 73 , 249–254. 26. Fassina, G ., Corti, A. & Cassani, G. (1992) Affinity enhancement of comple mentary p ep tide re cognitio n. Int. J. Pept. Prot. Res. 39, 549–556. 27. Tam, J .P. (1996) Recent advances in multiple antigen p eptides. J. Immun. Methods 196, 17–32. 28. Nakamura, T., Furanaka, H., Miyata, T., Tokunaga, F., Muta, T. & Iwanaga, S. (1988) T achyplesin, a class of a ntimicrobial peptide from the Horseshoe Crab (Tachypleustridentatus). J. Biol. C hem . 263, 16709–16713. 29. Kokryakov, V.N., Harwig, S.S., Pan yutich, E.A., Shevchenko, A.A., Aleshina, G.M., Shamova, O.V., Korneva, H.A. & Lehrer, R. (1993) Proteg rins: le ukocyte antimicrobial peptides that com- bine features of corticostatic defensins and tachyplesins. FEBS Lett. 327, 231–236. 30. T an g, Y Q., Yu a n, J., O ¨ sapay, G., O ¨ sapay, K., Tran, D., M iller, C.J., Ouellett e, A .J. & Selsted, M.E. ( 1999) A cyclic antimicro bial peptide produced in primate leukocytes by the ligation of two truncated alpha-defensins. Science 286, 498–502. 31. Castro, B., Dormay, J.R., Evin, G . & S elve, C. (1975) Reactifs de couplage peptidique IV (1)-L¢hexafluorophosphate de benzotria- zilyl N-oxytrisdimethlamino phosphonium (BOP). Tetra. Lett. 14, 1219–1222. 32. Lu, Y A. & Felix, A.M. (1993) Pegylated peptides I: solid-phase synthesis of N a -pegylated peptides using F moc strategy. Pept. Res. 6, 140–146. 33. L ehrer, R.I., Rosenm an, M., Harwing, S.S.L., Ja ckson, R . & Eisenhauer, P. (1991) Ultrasensitive assays for endogenous antimicrobial polypeptides. J. Immunol. Med. 137, 167–173. 34. Fehlbaum, P., Bulet, P ., Machaut, L., Lagueux, M., Broekaert, W.F., Hetru, C. & Hoffmann, J.A. (1994) Insect immunity, septic injury of Drosophila indu ces the synth esisof a potent antifungal peptide with sequence homology to plant antifungal peptides. J. Biol. C hem. 269, 33159–33163. 35. Kawano, K., Yoneya, T., Miyata, T., Yoshikawa, K., Tokunaga, T., Terada, Y. & Iwanaga, S. (1990) Anti microbial peptide, tachyplesin I, isola ted fro m hemoc ytes of the h orse shoecrab (Tachypleus tridentatus). J. Biol. Chem. 265, 15365–15367. 36. Yonezawa, A., Kuwahara, J., Fujii, N. & Sugiura, Y. (1992) Binding of tac hyplesin-I to DN A revealed b y footprinting analysis-significant contribution of secondary structure to DNA- binding and implication f or biological action. Biochemistry 31 , 2998–3004. Ó FEBS 2002 Antimicrobial dendrimeric peptides (Eur. J. Biochem. 269) 931 37. Fahrne r, R.L., Dieckmann, T., Harwig, S.S.L., L ehrer, R.I., Eisenberg, D. & Feigon, J. (1996) Solution structure of protegrin-1, a broad-spectrum antimicrobial peptide from porcine leukocytes. Chem. Biol. 3, 543–550. 38. Merrifield, R.B. (1986) Solid phase peptide synthesis. Science 57, 1619–1625. 39. Bals,R.,Wang,X.,Zasloff,M.&Wilson,J.M.(1998)Thepeptide antibiotic LL-37/hCAP-18 is expressed in epithelia of th e human lung whe re it has broad antimicrobial activity at the airway sur- face. J. Proc. Natl Acad. Sci. USA 95, 9541–9546. 40. Wadhwa, M.S., Collard, W.T., A dami, R .C. & Rice, K.G. (1997) Peptide-mediated gene delivery. influence of peptide structure on gene expression. Bioconj. Chem. 8, 81–88. 41. Pillai, O., Nair, V., Poduri, R. & Panchagnula, R. (1999) Trans- dermal iontophoresis. Part II: peptide and protein delivery. Methods Find. Expe. Clin. Pharm. 21, 229–240. 42. Bernkop-Schnurch, A. & Pasta, M. (1998) Intestinal peptide and protein delivery: novel bioadhesive drug-c arrier matrix shielding from enzymatic attack. J. Pharm. Sci. 87, 430–434. 43. Murphy, J.E., Uno, T., Hamer, J.D., Cohen, F.E., Dwarki, V. & Zuckermann, R.N . (1998) A combinatorial approach to th e dis- covery of efficient c ationic p eptoid reagents fo r g ene d elivery. Proc. Natl Acad. Sci. USA 95, 1517–1522. 44. Sheldon, K., Liu, D., Ferguson, J. & G ariepy, J. (1995) Loligo- mers. D esign of de n ovope ptide-based intracellular veh icles. Proc. Natl Acad. Sci. USA 92, 2056–2060. 932 J. P. Tam et al. (Eur. J. Biochem. 269) Ó FEBS 2002 . displayed activity spectra and potency comparable to naturally occurring antimicrobial peptides such as tachyplesins and protegrins. Dendrimeric RLYR (R4). N-hydroxybenzotriazole (HOBt), N,N¢-dicyclohexyl carbodiimide (DCC), and p-[(R,S)-a-[1-(9H-Fluoren-9-yl)-methoxylformamido]-2,4- dimethoxylbenzyl]-phenoxyacetic