A conformation-constrained peptide library based on insect defensin A

Peptides 25 (2004) 629–635

A conformation-constrained peptide library based on insect defensin A An Zhao, Yanning Xue∗ , Jie Zhang, Bo Gao, Jiannan Feng, Canquan Mao, Li Zheng, Nongle Liu, Fang Wang, Huixin Wang Beijing Institute of Basic Medical Sciences, P.O. Box 130 (3), Beijing 100850, PR China Received 8 May 2003; accepted 20 January 2004

Abstract Here, we reported a conformation-constrained peptide library, that was constructed based on the scaffold of a 29 amino acids peptide derived from insect defensin A. The peptide scaffold was designed utilizing the InsightII molecular modeling software and then displayed on M13 filamentous bacteriophage by fusion with coat protein III. The library was constructed by randomization of seven positions located within the two loops of the peptide scaffold generating approximately 8.3 × 108 transformants. Sequences from 14 randomly selected phage clones indicated that the distribution of nucleotides and amino acids paralleled with the expected frequency. Screening against the target proteins: tumor necrosis factor ␣, TNF receptor 1, TNF receptor 2 and monoclonal antibody against BMP-2 showed significant enrichment in all cases. The results presented here show that the reconstructed insect defensin A domain will be a promising non-antibody protein scaffold for the presentation of a phage-displayed constrained peptide library. © 2004 Elsevier Inc. All rights reserved. Keywords: Conformation-constrained peptide library; Insect defensin A; Screening

1. Introduction Phage display, based on the expression of the recombinant peptide or protein as a fusion with a coated protein of bacteriophage, represents a powerful new method for generating peptides and proteins that have novel-binding activities. The physical linkage between each displayed molecule and the DNA encoding it allows rapid identification of target-binding sequences by an in vitro selection process [1]. The ability to create a large repertoire of variants and to select the molecules with desired properties has made this technology applicable to a wide range of problems. Since the first phage library was introduced, which consisted of a peptide with six random residues fused to the N-terminal of coated protein pIII of phage [19], many different peptide libraries have been built and successfully used for the selection of novel ligands that bind to antibodies, cell surface receptors, DNA and many other targets [4]. The most commonly used peptide libraries are constructed as linear sequences or are constrained by disulfide bonds. Recently, other protein domains such as lipocalin [17,18], CTLA-4 ∗

Corresponding author. Tel.: +86-10-68163140; fax: +86-10-68213039. E-mail address: [email protected] (Y. Xue).

0196-9781/$ – see front matter © 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.peptides.2004.01.022

[8,13], CBD [20] and single protein A [12] have been investigated for use as scaffolds in conformation-constrained libraries. With the three-dimensional structure of scaffold protein maintained, a conformation-constrained library can be constructed by grafting the randomized sequences onto the molecular surface. From the population of different, though structurally related, phage displayed peptides, those members with desired binding properties can be isolated by affinity selection. Compared with the more flexible form, conformational ligand might be expected to have a higher affinity, due to the smaller loss of entropy upon binding; meanwhile, the topological structure of selected ligand has been predefined by scaffold protein [14], facilitating the interpretation of the molecular interaction at atomic level. A conformation-constrained peptide library might thus be an ideal tool for structure-based ligands selection, for use in many biological and medical applications especially in drug discovery. Insect defensin A is a small protease-resistant molecule with the follow peptide sequence ATCDLLSGTGINHSACAAHCLLRGNRGGYCNGKGVCVCRN. In 1995, its three-dimensional structure in solution was determined by NMR [3], and was found to contain an ␣-helix and a twisted antiparallel ␤-sheet folded into a stable structure by three

630

A. Zhao et al. / Peptides 25 (2004) 629–635

disulfide bonds. Two interval loops, directed to the surface, are predicted to be tolerant for extensive mutagenesis, suggesting that insect defensin would be a good candidate for use in scaffold engineering. Previous biological tests suggest however, that the natural occurring insect defensin A is difficult to display on the surface of phage, as it has been reported to participate in innate immunity and exhibit a wide spectra of antimicrobial activity [6]. In this study, we attempted to acquire a 29 amino acids reconstructed molecule on the basis of three-dimensional structure of insect defensin A and explore its potential for use as protein scaffold in the construction of a random phage peptide library. The functionality of this peptide library was tested by using the monoclonal antibody against BMP-2 (McAbBMP-2), tumor necrosis factor ␣ (TNF␣), TNF receptor 1 (TNFR1) and TNF receptor 2 (TNFR2) as target proteins in selection of binding peptide.

2. Materials and methods 2.1. Strains and vectors Escherichia coli strain TG1 K12 (lac-pro), supE, thi, hsd5/F [traD36, proAB, lacIq , lacZ  M15]was used as a host during library construction and for the phage work. E coli strain HB2151 was used for protein production. The phagemid pCANTAB 5E was used for the library construction [11]. 2.2. Modelling of insect defensin A Three-dimensional structures of the native and reconstructed defensin A were modeled using Homology Method

(Insight II software, MSI Inc.) based on the native crystal structure with an Octane2 graphics terminal. In order to minimize steric clashes and ensure correct bond lengths and angles after modeling, the modeled structure was optimized and subjected to 3000 steps of energy minimization. To ensure that the conformation of the modeled structure was not trapped in a local potential energy minimum, the side-chain positions of the residues in the loop was subjected to simulated heating and molecular dynamics at elevated temperatures followed by slow cooling to a low energy conformation. All molecular mechanism calculations were dealing with CVFF force field on SGI workstation. 2.3. Construction of reconstructed insect defensin Synthetic oligonucleotides were used to generate the gene encoding reconstructed insect defensin A, in which seven glycines coded for by GGA or GGT were introduced into two loop regions, as shown in Fig. 1. Two overlapping oligonucleotides coding for a 29 amino acids peptide were synthesized, with the following sequences 5 -GTCCTCGCAACTGCGGCCCAGCCGGCCAACATCTCCGCTTGCGCTGCAATCTGCCTGCTGGGAGGTGGAGGTGGTGGTTACGCTAAC-3 , and 5 -GAGTCATTCTCGACTTGCGGCCGCGTTAGTGCATACGCAAACACCTCCACCGTTAGCGTAACCACC-3 . NNK or nnk represent either GGA or GGT substitutions within the two loop regions. These two overlapping oligonucleotides were then annealled by first incubating at 94 ◦ C and then slowly cooling to below 37 ◦ C. Klenow DNA polymerase was used to generate the complementary strand by incubating at 37 ◦ C for 30 min. Following inactivation of the Klenow DNA polymerase, the double-stranded DNA was digested with SfiI (5 ) and NotI (3 ) and ligated into the phagemid vector pCANTAB 5E [11].

5 -GTC CTC GCA ACT GCG GCC CAG CCG GCC AAC ATC TCC GCT TGC GCT 3’-cag gag cgt tga cgc cgg gtc ggc cgg ttg tag agg cga acg cga N

I

S

A

C

A

GCA ATC TGC CTG CTG NNK NNK NNK NNK GGT GGT TAC GCT AAC nnk nnk cgt tag acg gac gac nnm nnm nnm nnm CCA CCA ATG CGA TTG NNM NNM A

I

C

L

L

X

X

X

X

G

G

Y

A

N

X

X

nnk gtt tgc gta tgc act aac gcg gcc gca agt cga gaa tga ctc -3 NNM CAA ACG CAT ACG TGA TTG CGC CGG CGT TCA GCT CTT ACT GAG –5’ X

V

C

V

C

T

N

Fig. 1. A schematic diagram for library construction. Two overlap oligonucleotides are shown in bold and capital letter, whereas complementary nucleotides are shown in lower case. The encoded amino acids are shown in single capital letter. The restriction site for SfiI and NotI are underlined at 5 and 3 ends respectively. Seven randomized residues in two loop region are boxed, where N is G, C, A or T; K is G or T; M is C or A and X are variable amino acid residues.

A. Zhao et al. / Peptides 25 (2004) 629–635

2.4. Construction of the library Gene pool encoding peptide library members with seven random amino acids were also generated by two complementary oligonucleotides. Based on the DNA sequence of the 29 amino acid peptide, two overlapping oligonucleotides containing seven random and degenerate codons at the corresponding position were designed and synthesized, as shown in Fig. 1. The oligonucleotide sequences are: 5 -GTCCTCGCAACTGCGGCCCAGCCGGCCAACATCTCCGCTTGCGCTGCAATCTGCCTGCTG(NNK)4 GGTGGTTACGCTAAC-3 , and 5 -GAGTCATTCTCGACTTGCGGCCGC GTTAGTGCATACGCAAAC (MNN)3 GTTAGCGTAACCACC-3 . After annealing, complementation and digestion, the DNA fragment containing SfiI and NotI at 5 and 3 ends respectively was cloned into pCANTAB 5E vector as described above. 2.5. E. coli transformation For display on the surface of M13 bacteriophage, DNA constructs were electroporated into the electrocompetent E. coli strain TG1 using standard procedures. 2.6. Preparation of phage stock Approximately 1010 cells from the constructed library were added to 100 ml 2× YT medium containing 100 ␮g/ml of ampicillin and 2% glucose and grown with shaking at 37 ◦ C until the OD600 measured 0.5. The cells were infected at a ratio of 10 with helper phage M13KO7: 1 bacteria by incubating at 37 ◦ C for 1 h. The superinfected cells were spun down and resuspended in 2× YT medium containing 100 ␮g/ml of ampicillin and 50 ␮g/ml of kanamycin. Following overnight incubation at 30 ◦ C with vigorous shaking, the phage particles were harvested by centrifugation and PEG precipitation as described. Harvested phage were finally dissolved in PBS and titrated for biopanning. 2.7. Biopanning Selections was performed according to previously published protocols with the following modifications [12]. Target proteins were coated onto ELISA microtiter plate at 10 ␮g/well, and wells blocked with 10 mg/ml BSA in PBS and pre-washed with PBST (PBS with 0.05% Tween 20). Approximately 100 ␮l of the pre-treated phage stock containing 1013 phage particles was then added to the wells and incubated for 2 h at room temperature. Thereafter the wells were washed twice with PBST (PBS with 0.1% Tween 20) in the first panning cycle, four times in the second cycle, five times in the third cycle and six times in the fourth cycle. Phage particles were eluted with 100 ␮l of glycine buffer (pH2.2) by incubating at room temperature for 8 min. The eluted phage particles were then neutralized by adding

631

15 ␮l of 1 M Tris–HCl (pH 9.1) solution and were used for the titration procedure and new phage stock preparation. Enrichment was monitored by the ratio between titer from target-coated well (P-value) and titer from BSA-coated well (N-value). 2.8. Identification of selected bacterial clones After three or four rounds of biopanning some bacterial clones were randomly picked from the enriched library. Phagemid was purified for sequencing as described in manufacturer’s manual. Sequencing of the insert was performed by Bioasia Inc. (Shanghai, China).

3. Results 3.1. Design of insect defensin A In order to acquire a suitable scaffold for designing a conformation-constrained peptide library, the insect defensin A was rationally designed for display on the surface of the phage. Based on the structural consideration, a fragment of 29 amino acids at the C-terminal of definsins was obtained from truncating a region out with secondary structure. Four amino acid residues were chosen for substitution in order to eliminate antimicrobial activity (His-13 to Ile, His-19 to Ile and Arg-39 to Thr) [3], and one residue was changed for production in a bacterial expression system (Cys-30 to Ala). Following the rational design, the final structure of the molecule was predicted and analyzed by the Insight II molecular modeling software. As illustrated in Fig. 2, the molecule was predicted to hold the original structure. The newly reconstructed molecule consisted of a 29-amino-acid peptide, forming an ␣-helix and two ␤-strands, stabilized by two disulfide bridges. 3.2. Phage display of the reconstructed molecule To investigate if the 29 amino acid peptide could be displayed on the filamentous phage, a DNA fragment encoding the reconstructed molecule was synthesized and inserted into the pCANTAB 5E at the cloning sites of Sfi I and Not I. The resultant plasmid, named pCANTAB 5E-1ICA29 ,was then transformed into E. coli TG1. Upon superinfection with helper phage M13KO7, the pCANTAB 5E proteins including hybrid pIII-1ICA29 were expressed. The engineered recombinant protein contains an E-tag at C-terminus, allowing for detection by Western blot assay. As shown in Fig. 3, the recombinant phage coat protein g3p, but not the wild-type, reacted with anti-E-tag mAb, suggesting that the 1ICA29 -g3p fusion protein had been produced and displayed on the surface of the phage. Moreover, this vector also contained an amber codon between the E-tag and the start of pIII resulting in production of soluble displayed peptide in

632

A. Zhao et al. / Peptides 25 (2004) 629–635

Fig. 2. Tertiary structure of the insect defensin A (A). The polypeptide backbone is shown in a ribbon representation (Insight II molecular modeling software), N-terminal 11 amino acids is shown in fold line. Proposed Tertiary structure of the reconstructed molecule(B).The four substitution of His-13,19 → Ile, Arg-39 → Thr and Cys30 → Ala are shown in yellow.

non-suppression E. coli HB2151. The 1ICA29 was purified from the periplasmic space of E. coli HB2151 cultures harboring the pCANTAB 5E-1ICA29 construct. The result of SDS–PAGE analysis showed that the 1ICA29 protein had been produced and was of the expected size (3 kDa ) (Fig. 3). 3.3. Library Construction and characterization After the reconstructed molecule had been displayed, we utilized it as scaffold to construct the random peptide library. The two interval loops located in residues 23–26 and 32–34 (as shows in Fig. 1) of the molecular surface were used to introduce randomization. The DNA fragment corresponding to the 29 amino acid peptide with seven random amino acid was synthesized, and the random amino acid were encoded by the NNK (N = A, C, G or T, K = G or T) degeneracy.

Fig. 3. Analysis of expressed fusion protein. (A) 12% SDS–PAGE analysis of expressed recombinant phage. Lane 1: molecular weight marker, Lane 2: recombinant phage. (B) Western blot analysis of the fusion protein from the pCANTAB 5E phagemid by anti-E-tag mAb, Lane 3: the wild-type phage, Lane 4: the recombinant phage coat protein III. (C) 18% SDS–PAGE analysis of purified 1ICA29 , Lane 5: low molecular weight marker, lane6: 1ICA29 .

This prevented the occurrence of the non-suppressible stop codons TGA and TAA in the randomized segment, which would have resulted in noninfectious phage. After electroporation of E. coli with phagemid ligation mixtures, a library with approximately 8.3×108 independent transformants was obtained. A total of 33 clones picked from the library were subjected to PCR screening. The result showed that more than 97% (32/33) of the clones contained an insert of the expected length, suggesting the efficient ligation of DNA with the phagemid vector (data not shown). In addition, 14 clones randomly picked from native library were sequenced for the analysis of quality and heterogeneity of the library (as shown in Fig. 4). All clones contained degenerate codons corresponding to the library design in which the last nucleotide was designated to G or T, and further, the distribution of nucleotides and amino acids at the seven degenerate positions

NISACAAICLL NISACAAICLL NISACAAICLL NISACAAICLL NISACAAICLL NISACAAICLL NISACAAICLL NISACAAICLL NISACAAICLL NISACAAICLL NISACAAICLL NISACAAICLL NISACAAICLL NISACAAICLL

LTWA QGGM QFWT VLIS ADSR QRRV RWCA VWCR GGKR PGFA IYDQ AVKG RWLQ SLGE

GGYAN GGYAN GGYAN GGYAN GGYAN GGYAN GGYAN GGYAN GGYAN GGYAN GGYAN GGYAN GGYAN GGYAN

VVL QVT APA KPV KQS ARF VFL RNN GAA QSS RLF YQA PLI QRQ

VCVCTN VCVCTN VCVCTN VCVCTN VCVCTN VCVCTN VCVCTN VCVCTN VCVCTN VCVCTN VCVCTN VCVCTN VCVCTN VCVCTN

Fig. 4. Amino acid sequences from 14 randomly selected phage. Residues included in the randomization are boxed.

A. Zhao et al. / Peptides 25 (2004) 629–635

633

Table 1 Nucleotide distribution in the variable region of 14 randomly selected infectious phage

A G C T

Frequency of each base by position in codon (%)a

45

1

2

3

40

14.29 31.63 22.45 31.63

22.45 25.51 27.55 24.49

– 59.18 – 40.81

35 30 25

a The frequency = base in certain position/98 bases(14 clones × 7 codons) × 100%.

P/N

Base

1st round 2nd round 3rd round 4th round

20 15 10

were almost identical with the expected frequency (Tables 1 and 2). Result revealed that the library had sufficient complexity and diversity and thus could be used for the screening of peptide ligands. 3.4. Selection against TNFα, TNFR1, TNFR2, McAbBMP-2 To test the utility of the insect defensin A scaffold library, four proteins with different size and origin were chosen as target to perform screening: TNF␣, TNF receptor 1, TNF receptor 2 and monoclonal antibody against BMP-2. Four rounds of biopanning with the above library was carried out against TNF␣ and McAbBMP-2, in parallel with

Table 2 Amino acid distribution in the variable region of 14 randomly selected infectious phage Amino acid

Codons

Expected frequencya (%)

Observed frequencyb (%)

Observed/ Expected

Arg Leu Ser Ala Gly Pro Thr Gln Val Asn Asp Cys Glu His Ile Lys Met Phe Trp Tyr

CGK, AGG CTK, TTG TCK, AGT GCK GGK CCK ACK CAG, TAGc GTK AAT GAT TGT GAG CAT ATT AAG ATG TTT TGG TAT

9.4 9.4 9.4 6.3 6.3 6.3 6.3 6.3 6.3 3.1 3.1 3.1 3.1 3.1 3.1 3.1 3.1 3.1 3.1 3.1

11.22 8.16 6.12 10.20 8.16 4.08 3.06 11.22 9.18 2.04 2.04 2.04 1.02 0 3.06 4.08 1.02 5.10 5.10 2.04

1.19 0.86 0.65 1.61 1.29 0.64 0.50 1.78 1.45 0.65 0.65 0.65 0.32 0 0.98 1.31 0.32 1.64 1.64 0.65

a Expected frequency = codons for that amino acid/32 codons × 100%. b Observed frequency = codons for that amino acid observed/98 sequenced codons × 100%. c The stop codon TAG is suppressed by Gln in the strain used to propagate the library.

5 0

TNF

McAbBMP-2

TNFR1

TNFR2

target proteins Fig. 5. Enrichment of the phage that bind to TNF␣, McAbBMP-2, TNFR1 and TNFR2 are shown respectively. Enrichment was determined as the ratio between eluted phage from target protein (P value) coated well and eluted phage from BSA coated well (N value).

BSA as a negative control. A final of 24- and 45-fold enrichments were yielded, respectively, in the ratio of eluted phages between positive and negative. And for the target TNFR1 and TNFR2, the similar enrichments were also observed after three rounds of panning (Fig. 5). To confirm if the library described here could be used for identifying specific peptides to target protein, fourteen clones were picked from the enriched library after three rounds of biopanning against TNFR2, and their DNA were isolated for sequencing. According to the sequencing result, twelve different peptides structures were obtained. These peptides showed significant consensus in sequences and conservation in amino acid composition (Fig. 6), suggesting that the TNFR2-binding peptides with biased sequence has been obtained by selection from the library. Loop 1

Loop2

HREY-----NHS LRQT-----WVD DRVL-----WYQ DRIS-----YRQ EKFR-----RWK ARYT-----VPR KRWG-----FHE RSHL-----IRQ RQGL-----RGY RQWV-----RLN REFR-----WPR RHRM-----PTY

frequency

1 2 1 1 1 1 2 1 1 1 1 1

Fig. 6. Sequence alignment of selected TNFR2-binding peptides. Amino acid residues only in two loops were shown. The biased amino acids were specified in bold. The number after each sequence indicates the frequency among the selected clones.

634

A. Zhao et al. / Peptides 25 (2004) 629–635

4. Discussion In this work, we reported a new scaffold of a 29 amino acids peptide derived from insect defensin A for the construction of a conformation-constrained peptide library. Insect defensin A was chosen as a scaffold in view of several structural features: small size, stable structure and two loop regions with tolerance for insertion [3]. However, biological characterization suggested that insect defensin A serves as an effector molecule in innate immunity, providing an efficient initial defense against a wide range of bacteria and fungi [6]. If the insect defensin A was to be used as a scaffold to generate a library, it was necessary to eliminate the antimicrobial activity. The primary sequence of insect defensin A contains many hydrophobic and positively charged residues, that interact with bacteria membrane, suggesting a possible mechanism for its antimicrobial effects [15]. Indeed, previous biological tests suggest that the bacterial cytoplasmic membrane is the target of defensin A. The defensin A – lipid complexes have been observed in phospholipid monolayers exposed to insect defensin and these complexes are non miscible in the lipid phase and induce microheterogeneity in the lipid membrane [10]. This lead us to propose that the substitution of basic amino acid residues located at the middle portion of the sequence may alter the microbicidal potency [3]. Increasing evidence supports that the antimicrobial activity of defensins could vary according to the primary sequences and the amphiphilic structure [21]. Simple analogs that mimic structure of bovine neutrophil ␤-defensin obtained by the Raj P.A. group was found to elicit microbicidal potency comparable to native defensins [7,16]. Although there are differences in the sequence and folding patterns between mammalian and insect defensin A, antimicrobial activity might be eliminated with the destruction of the amphiphilic structure. Utilizing Insight IImolecular modeling software, the definsin A was reconstructed and was shown to hold the original tertiary structure. In construction of the library, the loop region was found to be an almost exclusive choice to introduce randomization, rare among secondary structure element such as ␣-helix and ␤-sheet [5]. It is possible that the intrinsic flexibility of loop regions result in little or no effect on the substitution on protein folding. In our library, two loop regions instead of one were randomized, this could be advantageous for increasing both the diversity of the library and the complexity of the surface structures involved in intermolecular recognition [9]. In the conformation-constrained library, depending on the structure of the scaffold protein and the region chosen for randomization, different topologies of the interaction surface can be achieved, Such as a tip-like surface described for libraries of the CBD [12] and the more crevice-like surface as described for the library of lipocalin [2]. The library based on the insect defensin A as described here was constructed by randomized at two separate-binding sites on the flat surface, the resultant library could be a suitable resource for selection of ligands that bind with a larger surface area. However, the

length of the randomized region is only seven amino acids in total, which might restrict its affinity with the target. As the library is in conformation-constrained format, determination of the tertiary structure of displayed molecule is very important. It is reported that the NMR and Mass Spectrometry will be very useful for this requirement, however sufficient quantities of soluble material are necessary, however, in this instance we failed to obtain enough for detection. Even so, significant enrichments were observed in all screening cases, indicating the scaffold used to construct the library might be in a stable structure. Further evidence that the scaffold was correct came from the sequence of selected peptides. After three rounds of panning against TNFR2 we randomly picked infected bacterial colonies for sequencing the displayed peptides. Significant bias in amino acid composition can be observed in the variable loop region of the library (Fig. 6). The first amino acid residue in loop 1 tended to be either a basic amino acids (6 out of 12 sequences) or an acidic amino acids (3 out of 12), however, when the first amino acid was acidic the second amino acid was always basic, implying that the basic amino acid was the more important residue in this region for the binding to TNFR2. Other biases include aromatic amino acid at the third and the fifth residues (both 5 out of 12), basic amino acids at the fifth and seventh position (both 3 out of 12) and acyl amino acid at the seventh position (4 out of 12). We could not find homologous structure in the ligand of TNFR2, such as TNF␣ or TNF␤, possibly because the seven residues in the loop region were separated in steric structure. In conclusion, we report here a new conformationconstrained peptide library, whereby the library was constructed on the scaffold of insect defensin A and has sufficient complexity and diversity for peptide ligand screening. The library may serve as an important tool in finding new ligands for the target protein.

Acknowledgments We thank Dr. Alison Engles at Edingburgh University for suggestions and revisions on the manuscript. We are grateful to all the members in our laboratory for their help. This work was supported by a National High-Tech Science Fundation of China (No. 2002AA233101).

References [1] Benhar I. Biotechnological applications of phage and cell display. Biotechnol Adv 2001;19:1–33. [2] Beste G, Schmidt FS, Stibora T, Skerra A. Small antibody-like proteins with prescribed ligand specificities derived from the lipocalin fold. Proc Natl Acad Sci USA 1999;96:1898–903. [3] Cornet B, Bonmatin JM, Hetru C, Hoffmann JA, Ptak M, Vovelle F. Refined three-dimensional solution structure of insect defensin A. Structure 1995;3:435–48.

A. Zhao et al. / Peptides 25 (2004) 629–635 [4] Cortese R, Monaci P, Luzzago A, Santini C, Bartoli F, Cortese I, et al. Selection of biologically active peptides by phage display of random peptide libraries. Curr Opin Biotechnol 1996;7:616–21. [5] Doi N, Yanagawa H. Screening of conformationally constrained random polypeptide libraries displayed on a protein scaffold. Cell Mol Life Sci 1998;54:394–404. [6] Hoffmann JA, Kafatos FC, Janeway CA, Ezekowitz RA. Phylogenetic perspectives in innate immunity. Science 1999;284:1313–8. [7] Hoover DM, Rajashankar KR, Blumenthal R, Puri A, Oppenheim JJ, Chertov O, et al. The structure of human beta-defensin-2 shows evidence of higher order oligomerization. J Biol Chem 2000; 275:32911–8. [8] Hufton SE, van Neer N, van den Beuken T, Desmet J, Sablon E, Hoogenboom HR. Development and application of cytotoxic T lymphocyte-associated antigen 4 as a protein scaffold for the generation of novel binding ligands. FEBS Lett 2000;475: 225–31. [9] Lander RC, Ley CA. Novel frameworks as a source of high-affinity ligands. Curr Opin Biotech 2001;12:406–10. [10] Maget-Dana R, Ptak M. Penetration of the insect defensin A into phospholipid monolayers and formation of defensin A-lipid complexes. Biophys J 1997;73:2527–33. [11] McCafferty J, Fitzgerald KJ, Earnshaw J, Chiswell DJ, Link J, Smith R, et al. Selection and rapid purification of murine antibody fragments that bind a transition-state analog by phage display. Appl Biochem Biotechnol 1994;47:157–71, discussion 171–3. [12] Nord K, Gunneriusson E, Ringdahl J, Stahl S, Uhlen M, Nygren PA. Binding proteins selected from combinatorial libraries of an

[13]

[14] [15] [16]

[17]

[18] [19]

[20]

[21]

635

alpha-helical bacterial receptor domain. Nat Biotechnol 1997;15: 772–7. Nuttall SD, Rousch MJ, Irving RA, Hufton SE, Hoogenboom HR, Hudson PJ. Design and expression of soluble CTLA-4 variable domain as a scaffold for the display of functional polypeptides. Proteins 1999;36:217–27. Nygren PA, Uhlen M. Scaffolds for engineering novel binding sites in proteins. Curr Opin Struct Biol 1997;7:463–9. Raj PA, Dentino AR. Current status of defensins and their role in innate and adaptive immunity. FEMS Microbiol Lett 2002;206:9–18. Raj PA, Karunakaran T, Sukumaran DK. Synthesis, microbicidal activity, and solution structure of the dodecapeptide from bovine neutrophils. Biopolymers 2000;53:281–92. Schlehuber S, Beste G, Skerra A. A novel type of receptor protein, based on the lipocalin scaffold, with specificity for digoxigenin. J Mol Biol 2000;297:1105–20. Skerra A. Lipocalins as a scaffold. Biochim Biophys Acta 2000; 1482:337–50. Smith GP. Filamentous fusion phage: novel expression vectors that display cloned antigens on the virion surface. Science 1985; 228:1315–7. Smith GP, Patel SU, Windass JD, Thornton JM, Winter G, Griffiths AD. Small binding proteins selected from a combinatorial repertoire of knottins displayed on phage. J Mol Biol 1998;277:317–32. Wu M, Maier E, Benz R, Hancock RE. Mechanism of interaction of different classes of cationic antimicrobial peptides with planar bilayers and with the cytoplasmic membrane of Escherichia coli. Biochemistry 1999;38:7235–42.