Purification of polyclonal anti-conformational antibodies for use in affinity selection from random peptide phage display libraries: A study using the hydatid vaccine EG95

The use of polyclonal antibodies to screen random peptide phage display libraries often results in the recognition of a large number of peptides that mimic linear epitopes on various proteins. There appears to be a bias in the use of this technology toward the selection of peptides that mimic linear epitopes. In many circumstances the correct folding of a protein immunogen is required for conferring protection. The use of random peptide phage display libraries to identify peptide mimics of conformational epitopes in these cases requires a strategy for overcoming this bias. Conformational epitopes on the hydatid vaccine EG95 have been shown to result in protective immunity in sheep, whereas linear epitopes are not protective. In this paper we describe a strategy that results in the purification of polyclonal antibodies directed against conformational epitopes while eliminating antibodies directed against linear epitopes. These affinity purified antibodies were then used to select a peptide from a random peptide phage display library that has the capacity to mimic conformational epitopes on EG95. This peptide was subsequently used to affinity purify monospecific antibodies against EG95.     

Split-intein mediated circular ligation used in the synthesis of cyclic peptide libraries in E. coli

Recent advances in chemical biology and the advantages presented by in vivo screening have highlighted the need for a robust and flexible biologically synthesized small-molecule library. Herein we describe a method for the biosynthesis of cyclic peptide libraries of up to 10(8) members in Escherichia coli using split-intein circular ligation of peptides and proteins (SICLOPPS). The method utilizes split-intein chemistry to cyclize randomized peptide sequences. The cyclic peptide library can potentially be of any size and the peptide itself may contain unlimited random residues. However, the library size is limited by the transformation efficiency of E. coli and random residues are generally limited to five, but additional amino acids can be used in the cyclic peptide backbone, varying the structure and ring size of the cyclic peptide. SICLOPPS libraries have been combined with a bacterial reverse two-hybrid system in our labs and used in the identification of inhibitors of several protein-protein interactions. This protocol is expected to take around 3-4 weeks to implement.     

Quantitative assessment of peptide sequence diversity in M13 combinatorial peptide phage display libraries

Novel statistical methods have been developed and used to quantitate and annotate the sequence diversity within combinatorial peptide libraries on the basis of small numbers (1-200) of sequences selected at random from commercially available M13 p3-based phage display libraries. These libraries behave statistically as though they correspond to populations containing roughly 4.0+/-1.6% of the random dodecapeptides and 7.9+/-2.6% of the random constrained heptapeptides that are theoretically possible within the phage populations. Analysis of amino acid residue occurrence patterns shows no demonstrable influence on sequence censorship by Escherichia coli tRNA isoacceptor profiles or either overall codon or Class II codon usage patterns, suggesting no metabolic constraints on recombinant p3 synthesis. There is an overall depression in the occurrence of cysteine, arginine and glycine residues and an overabundance of proline, threonine and histidine residues. The majority of position-dependent amino acid sequence bias is clustered at three positions within the inserted peptides of the dodecapeptide library, +1, +3 and +12 downstream from the signal peptidase cleavage site. Conformational tendency measures of the peptides indicate a significant preference for inserts favoring a beta-turn conformation. The observed protein sequence limitations can primarily be attributed to genetic codon degeneracy and signal peptidase cleavage preferences. These data suggest that for applications in which maximal sequence diversity is essential, such as epitope mapping or novel receptor identification, combinatorial peptide libraries should be constructed using codon-corrected trinucleotide cassettes within vector-host systems designed to minimize morphogenesis-related censorship.     

Mapping epitopes of neutralizing monoclonal antibodies using phage random peptide libraries

Identification of protective determinants from microbial proteins is a necessary step in the rational design of subunit vaccines. We have previously used a synthetic peptide scan (Pepscan) assay to map a panel of eight neutralizing monoclonal antibodies (mAb; designated as C1.1 to C1.8) to a common motif sequence from Chlamydia trachomatis. In the present study, five of the eight mAbs were used to screen phage random peptide libraries. mAbs C1.1 and C1.3 selected a motif sequence of G-L-X-N-D from a pIII-based phage random peptide library and a motif sequence of G-X-X-N-D from a pVIII-based random peptide library while mAbs C1.6 to C1.8 failed to select recognizable motifs from either of the phage libraries. However, C1.6 to C1.8 bound to the same motif sequence displayed on phage when the appropriate conformational constraints were imposed onto the motif sequence. Thus the specificity of the mAbs identified on Pepscan assays correlates with the mAbs’ dependence on local epitope constraints displayed on the phage surface. Received 12 August 1996/ Accepted in revised form 03 November 1996     

Isolation of a highly specific ligand for the alpha 5 beta 1 integrin from a phage display library

Our previous studies showed that the alpha 5 beta 1 integrin selects cysteine pair-containing RGD peptides from a phage display library based on a random hexapeptide. We have therefore searched for more selective peptides for this integrin using a larger phage display library, where heptapeptides are flanked by cysteine residues, thus making the inserts potentially cyclic. Most of the phage sequences that bound to alpha 5 beta 1 (69 of 125) contained the RGD motif. Some of the heptapeptides contained an NGR motif. As the NGR sequence occurs in the cell-binding region of the fibronectin molecule, this sequence could contribute to the specific recognition of fibronectin by alpha 5 beta 1. Selection for high affinity peptides for alpha 5 beta 1 surprisingly yielded a sequence RRETAWA that does not bear obvious resemblance to known integrin ligand sequences. The synthetic cyclic peptide GACRRETAWACGA (*CRRETAWAC*) was a potent inhibitor of alpha 5 beta 1-mediated cell attachment to fibronectin. This peptide is nearly specific for the alpha 5 beta 1 integrin, because much higher concentrations were needed to inhibit the alpha v beta 1 integrin, and there was no effect on alpha v beta 3- and alpha v beta 5-mediated cell attachment to vitronectin. The peptide also did not bind to the alpha IIb beta 3 integrin. *CRRETAWAC* appears to interact with the same or an overlapping binding site in alpha 5 beta 1 as RGD, because cell attachment to *CRRETAWAC* coated on plastic was divalent cation dependent and could be blocked by an RGD-containing peptide. These results reveal a novel binding specificity in the alpha 5 beta 1 integrin.     

A peptide isolated by phage display binds to ICAM-1 and inhibits binding to LFA-1

A mixed phage library containing random peptides from four to eight residues in length flanked by cysteine residues was screened using a recombinant soluble, form of human ICAM-1, which included residues 1–453, (ICAM-1_1–453). Phage bound to immobilized ICAM-1_1–453 were eluted by three methods: (1) soluble ICAM-1_1–453, (2) neutralizing murine monoclonal antibody, (anti-ICAM-1, M174F5B7), (3) acidic conditions. After three rounds of binding and elution, a single, unique ICAM-1 binding phage bearing the peptide EWCEYLGGYLRYCA was isolated; the identical phage was selected with each method of elution. Attempts to isolate phage from non-constrained (i.e., not containing cysteines) libraries did not yield a phage that bound to ICAM-1. Phage displaying EWCEYLGGYLRCYA bound to immobilized ICAM-1_1–453 and to ICAM-1_1–185, a recombinant ICAM-1, which contains only the two amino-terminal immunoglobulin domains residing within residues 1–185. This is the region of the ICAM-1 that is bound by LFA-1. The phage did not bind to proteins other than ICAM-1. The phage bound to two ICAM-1 mutants, which contained amino acid substitutions that dramatically decreased or eliminated the binding to LFA-1. Studies were also performed with the corresponding synthetic peptide. The linear form of the synthetic EWCEYLGGYLRCYA peptide was found to inhibit LFA-1 binding to immobilized ICAM-1_1–453 in a protein-protein binding assay. By contrast, the disulfide, cyclized, form of the peptide was inactive. The EWCEYL portion of the sequence is homologous to the EWPEYL sequence found within rhinovirus coat protein 14, a nonintegrin protein that binds to ICAM-1. Taken together, the results suggests that the EWCEYLGGYLRCYA sequence is capable to binding to immobilized ICAM-1. Phage display appears to represent a new approach for the identification of peptides that interfere with ICAM-1 binding to β2 integrins. © 1996 Wiley-Liss, Inc.     

Immunogenically fit subunit vaccine components via epitope discovery from natural peptide libraries

Antigenic peptides that bind pathogen-specific Abs are a potential source of subunit vaccine components. To be effective the peptides must be immunogenically fit: when used as immunogens they must elicit Abs that cross-react with native intact pathogen. In this study, antigenic peptides obtained from phage display libraries through epitope discovery were systematically examined for immunogenic fitness. Peptides selected from random peptide libraries, in which the phage-displayed peptides are encoded by synthetic degenerate oligonucleotides, had marginal immunogenic fitness. In contrast, 50% of the peptides selected from a natural peptide library, in which phage display segments of actual pathogen polypeptides, proved very successful. Epitope discovery from natural peptide libraries is a promising route to subunit vaccines.     

A peptide isolated from phage display libraries is a structural and functional mimic of an RGD-binding site on integrins

Many integrins recognize short RGD-containing amino acid sequences and such peptide sequences can be identified from phage libraries by panning with an integrin. Here, in a reverse strategy, we have used such libraries to isolate minimal receptor sequences that bind to fibronectin and RGD-containing fibronectin fragments in affinity panning. A predominant cyclic motif, *CWDDG/LWLC*, was obtained (the asterisks denote a potential disulfide bond). Studies using the purified phage and the corresponding synthetic cyclic peptides showed that *CWDDGWLC*-expressing phage binds specifically to fibronectin and to fibronectin fragments containing the RGD sequence. The binding did not require divalent cations and was inhibited by both RGD and *CWDDGWLC*-containing synthetic peptides. Conversely, RGD-expressing phage attached specifically to immobilized *CWDDGWLC*-peptide and the binding could be blocked by the respective synthetic peptides in solution. Moreover, fibronectin bound to a *CWDDGWLC*-peptide affinity column, and could be eluted with an RGD-containing peptide. The *CWDDGWLC*-peptide inhibited RGD-dependent cell attachment to fibronectin and vitronectin, but not to collagen. A region of the beta subunit of RGD-binding integrins that has been previously demonstrated to be involved in ligand binding includes a polypeptide stretch, KDDLW (in beta 3) similar to WDDG/LWL. Synthetic peptides corresponding to this region in beta 3 were found to bind RGD-displaying phage and conversion of its two aspartic residues into alanines greatly reduced the RGD binding. Polyclonal antibodies raised against the *CWDDGWLC*- peptide recognized beta 1 and beta 3 in immunoblots. These data indicate that the *CWDDGWLC*-peptide is a functional mimic of ligand binding sites of RGD-directed integrins, and that the structurally similar site in the integrin beta subunit is a binding site for RGD.     

Synthetic antibody libraries focused towards peptide ligands

Synthetic antibody libraries have proven immensely useful for the de novo isolation of antibodies without the need for animal immunization. Recently, focused libraries designed to recognize particular classes of ligands, such as haptens or proteins, have been employed to facilitate the selection of high-affinity antibodies. Focused libraries are built using V regions encoding combinations of canonical structures that resemble the structural features of antibodies that bind the desired class of ligands and sequence diversity is introduced at residues typically involved in recognition. Here we describe the generation and experimental validation of two different single-chain antibody variable fragment libraries that efficiently generate binders to peptides, a class of molecules that has proven to be a difficult target for antibody generation. First, a human anti-peptide library was constructed by diversifying a scaffold: the human variable heavy chain (V_H) germ line gene 3-23, which was fused to a variant of the human variable light chain (V_L) germ line gene A27, in which L1 was modified to encode the canonical structure found in anti-peptide antibodies. The sequence diversity was introduced into 3-23 (V_H) only, targeting for diversification residues commonly found in contact with protein and peptide antigens. Second, a murine library was generated using the antibody 26-10, which was initially isolated based on its affinity to the hapten digoxin, but also binds peptides and exhibits a canonical structure pattern typical of anti-peptide antibodies. Diversity was introduced in the V_H only using the profile of amino acids found at positions that frequently contact peptide antigens. Both libraries yielded binders to two model peptides, angiotensin and neuropeptide Y, following screening by solution phage panning. The mouse library yielded antibodies with affinities below 20 nM to both targets, although only the V_H had been subjected to diversification.     

T7 lytic phage-displayed peptide libraries exhibit less sequence bias than M13 filamentous phage-displayed peptide libraries

We investigated whether the T7 system of phage display could produce peptide libraries of greater diversity than the M13 system of phage display due to the differing processes of lytic and filamentous phage morphogenesis. Using a bioinformatics-assisted computational approach, collections of random peptide sequences obtained from a T7 12-mer library (X(12)) and a T7 7-mer disulfide-constrained library (CX(7)C) were analyzed and compared with peptide populations obtained from New England BioLabs' M13 Ph.D.-12 and Ph.D.-C7C libraries. Based on this analysis, peptide libraries constructed with the T7 system have fewer amino acid biases, increased peptide diversity, and more normal distributions of peptide net charge and hydropathy than the M13 libraries. The greater diversity of T7-displayed libraries provides a potential resource of novel binding peptides for new as well as previously studied molecular targets. To demonstrate their utility, several of the T7-displayed peptide libraries were screened for streptavidin- and neutravidin-binding phage. Novel binding motifs were identified for each protein.     

Development of an enzyme-linked immunosorbent assay for thiacloprid in soil and agro-products with phage-displayed peptide

A monoclonal antibody (3A5) that can recognize thiacloprid was produced, and a linear 8-residue peptide phage library was constructed. Six phage-displayed peptides were isolated from the linear 8-residue peptide phage library and a cyclic 8-residue peptide phage library. A phage enzyme-linked immunosorbent assay (ELISA) was developed to detect thiacloprid using a phage-displayed peptide. Under the optimal conditions, the half-maximal inhibition concentration (IC₅₀) and the limit of detection (IC₁₀) of the developed phage ELISA were 8.3 and 0.7 μg/L, respectively. Compared with the conventional ELISA, the sensitivity was improved more than 3-fold. The cross-reactivity (CR) was less than 0.08% for the tested structural analogues and was regarded as negligible. The recoveries of thiacloprid ranged from 80.3% to 116.3% in environmental and agricultural samples, which conformed to the requirements for residue detection. The amount of thiacloprid detected by phage ELISA in the samples was significantly correlated with that detected by high-performance liquid chromatography. The current study indicates that isolating phage-displayed peptides from phage display libraries is an alternative method for the development of a sensitive immunoassay and that the developed assay is a potentially useful tool for detecting thiacloprid in environmental and agricultural samples.     

Identification of peptide sequences that target to the brain using in vivo phage display

Phage display technology could provide a rapid means for the discovery of novel peptides. To find peptide ligands specific for the brain vascular receptors, we performed a modified phage display method. Phages were recovered from mice brain parenchyma after administrated with a random 7-mer peptide library intravenously. A longer circulation time was arranged according to the biodistributive brain/blood ratios of phage particles. Following sequential rounds of isolation, a number of phages were sequenced and a peptide sequence (CTSTSAPYC, denoted as PepC7) was identified. Clone 7-1, which encodes PepC7, exhibited translocation efficiency about 41-fold higher than the random library phage. Immunofluorescence analysis revealed that Clone 7-1 had a significant superiority on transport efficiency into the brain compared with native M13 phage. Clone 7-1 was inhibited from homing to the brain in a dose-dependent fashion when cyclic peptides of the same sequence were present in a competition assay. Interestingly, the linear peptide (ATSTSAPYA, Pep7) and a scrambled control peptide PepSC7 (CSPATSYTC) did not compete with the phage at the same tested concentration (0.2-200?pg). Labeled by Cy5.5, PepC7 exhibited significant brain-targeting capability in in vivo optical imaging analysis. The cyclic conformation of PepC7 formed by disulfide bond, and the correct structure itself play a critical role in maintaining the selectivity and affinity for the brain. In conclusion, PepC7 is a promising brain-target motif never been reported before and it could be applied to targeted drug delivery into the brain.     

Preferential selection of Cys‐constrained peptides from a random phage‐displayed library by anti‐glucitollysine antibodies

Phage‐displayed peptides recognized by two monoclonal antibodies against glucitollysine were selected. The most prominent feature of the peptide panel was the presence of paired Cys in most of them (21/24 peptides). The availability of a wide variety of peptides having differently spaced paired Cys, as well as truly linear Cys‐free peptides, gave the opportunity to explore the role of disulfide bridges in phage selection. Some Cys‐containing peptides came from a Cys‐flanked cyclic 9‐mer library, but most of them (18/21) were derived from a totally random 12‐mer library, and hence the presence of Cys was dictated by the selector antibodies. Motifs shared by several peptides (potentially involved in binding) often contained or were flanked by Cys residues. Binding of all Cys‐containing phage‐displayed peptides was abolished/decreased after a reducing treatment. Screening a random peptide library (without invariant Cys residues) is powerful enough to clearly reveal the need, preferences, and diversity of Cys‐mediated structural constraints for recognition. Copyright © 2008 European Peptide Society and John Wiley & Sons, Ltd.     

Synthesis and screening of a random dimeric peptide library using the one-bead-one-dimer combinatorial approach

Large combinatorial libraries of random peptides have been used for a variety of applications that include analysis of protein-protein interactions, epitope mapping, and drug targeting. The major obstacle in screening these libraries is the loss of specific but low affinity binding peptides during washing steps. Loss of these specific binders often results in isolation of peptides that bind nonspecifically to components used in the selection process. Previously, it has been demonstrated that dimerizing or multimerizing a peptide can remarkably improve its binding kinetics by 10- to 1000-fold due to an avidity effect. To take advantage of this observation, we constructed a random library of 12 amino acid dimeric peptides on polyethylene glycol acrylamide (PEGA) beads by modifying the 'one-bead-one-compound' approach. The chemical synthesis of 100,000 peptides as dimers can be problematic due to steric and aggregation effects and the presence of many peptide sequences that are difficult to synthesize. We have designed a method, described in detail here, to minimize the problems inherent in the synthesis of a dimeric library by modifying the existing 'split and pool' synthetic method. Using this approach the dimeric library was used to isolate a series of peptides that bound selectively to epithelial cancer cells. One peptide with the amino acid sequence QMARIPKRLARH bound as a dimer to prostate cancer cells spiked into the blood but did not bind to circulating hematopoeitic cells. The monomeric form of this peptide, however, did not bind well to the same LNCaP cell line. These data demonstrate that "hits" obtained from such a 'one-bead-one-dimer' library can be used directly for the final application or used as leads for construction of second generation libraries.     

Selection of peptides targeting the human sperm surface using random peptide phage display identify ligands homologous to ZP3

Analysis of the surface architecture of human spermatozoa is a necessary step in the development of new approaches to contraception and resolving the causes of human infertility. In this study we have utilized phage display technology to identify peptides that bind with high affinity to the surface of human spermatozoa. Fifteen- and twelve-mer random peptide phage display libraries were screened against paraformaldehyde-fixed spermatozoa and a number of sperm-binding peptides were identified. One peptide, M6, displayed a high level of affinity for the sperm surface and showed sequence homology with a dominant human ZP3 epitope (hZP 25-33). This peptide bound preferentially to the equatorial and post acrosomal domains of the sperm head and exhibited contraceptive activity by virtue of its capacity to impair the fusion of acrosome-reacted spermatozoa with the vitelline membrane of the oocyte. A similar form of contraceptive activity was also observed within an unrelated peptide, K6, derived from screening the 12-mer library. These results indicate that phage display technology is a powerful tool for developing reagents capable of targeting the human sperm surface, providing insights into the composition of this structure and the identity of targets susceptible to contraceptive attack and pathological disruption.     

Phage display screening of epithelial cell monolayers treated with EGTA: identification of peptide FDFWITP that modulates tight junction activity

Phage display was used to screen for peptides that modulate the activity of epithelial cell tight junctions. Panning with a phage library that displays random 7-mers was performed using monolayers of human bronchial epithelial cells (16HBE14o(-)) treated with a calcium chelator, ethylene glycol-bis(2-aminoethylether)- N, N, N', N'-tetraacetic acid (EGTA), to increase accessibility to the junctional complex/paracellular space, followed by subtractive panning. A novel peptide, FDFWITP, identified as a potential tight junction modulator, was synthesized in linear and cyclic forms with lysine residues added to improve solubility. The cyclic form of the peptide reduced transepithelial electrical resistance (TER) in a concentration-dependent manner (80% reduction at 100 microM and 95% reduction at 500 microM) and was reversible within 2 h; the linear form only affected TER at the highest concentration. Interestingly, the constrained peptide did not increase permeation of the model small molecule, fluorescein. The highly selective activity of FDFWITP supports the hypothesis that ions and small molecules may be transported paracellularly across tight junctions by separate pathways.     

Phage Peptide Libraries

Filamentous phage particles have been central in the construction of libraries displaying vast numbers of random peptides. These random peptides can be antigenically presented as fusions to coat proteins III and VIII of the phage. The isolation of ligate-reactive phage from an immense background of nonspecific phage is achieved by the biopanning process. Enrichment of reactive phage relative to unreactive phage occurs with alternate rounds of affinity selection to the desired molecular target and amplification of the specifically bound phage. This allows the isolation of rare binding species contained in the phage peptide libraries. Each phage particle contains the information in its genome pertaining to the type of random peptide insert displayed. Hence, the identification of binding motifs displayed on ligate-reactive phage is revealed by sequencing the relevant insert site in the phage genome. Phage peptide libraries have been used to isolate ligands to an array of protein ligates. The libraries have proved particularly effective in defining the binding sites of monoclonal antibodies and to some extent polyclonal sera. The analysis of the peptide insert sequences of a number of different clones of antibody binding phage can reveal a great deal about the nature and restriction of the amino acid residues critical for the antibody–antigen interaction.     

Selection of internalizing ligand-display phage using rolling circle amplification for phage recovery

Selection of phage libraries against complex living targets such as whole cells or organs can yield valuable targeting ligands without prior knowledge of the targeted receptor. Our previous studies have shown that noninfective multivalent ligand display phagemids internalize into mammalian cells more efficiently than their monovalent counterparts suggesting that cell-based selection of internalizing ligands might be improved using multivalently displayed peptides, antibodies or cDNAs. However, alternative methods of phage recovery are needed to select phage from noninfective libraries. To this end, we reasoned that rolling circle amplification (RCA) of phage DNA could be used to recover noninfective phage. In feasibility studies, we obtained up to 1.5 million-fold enrichment of internalizing EGF-targeted phage using RCA. When RCA was applied to a large random peptide library, eight distinct human prostate carcinoma cell-internalizing peptides were isolated within three selection rounds. These data establish RCA as an alternative to infection for phage recovery that can be used to identify peptides from noninfective phage display libraries or infective libraries under conditions where there is the potential for loss of phage infectivity.     

Increasing the activity of cell adherent cyclic NGR peptides by optimizing the peptide length and amino acid character

Cyclic peptides are an attractive modality for the development of therapeutics and the identification of functional cyclic peptides that contribute to novel drug development. The peptide array is one of the optimization methods for peptide sequences and also useful to understand sequence–function relationship of peptides. Cell adherent cyclic NGR peptide which selectively binds to the aminopeptidase N (APN or CD13) is known as an attractive tumor marker. In this study, we designed and screened a library of different length and an amino acid substitution library to identify stronger cell adhesion peptides and to reveal that the factor of higher binding between CD13 and optimized cyclic peptides. Additionally, we designed and evaluated 192 peptide libraries using eight representative amino acids to reduce the size of the library. Through these optimization steps of cyclic peptides, we identified 23 peptides that showed significantly higher cell adhesion activity than cKCNGRC, which was previously reported as a cell adhesion cyclic peptide. Among them, cCRHNGRARC showed the highest activity, that is, 1.65 times higher activity than cKCNGRC. An analysis of sequence and functional data showed that the rules which show higher cell adhesion activity for the three basic cyclic peptides (cCX₁HNGRHX₂C, cCX₁HNGRAX₂C, and cCX₁ANGRHX₂C) are related with the position of His residues and cationic amino acids.     

Biodistribution of filamentous phage peptide libraries in mice

In vivo phage display is a new approach to acquire peptide molecules that bind stably to a given target. Phage peptide display libraries have been selected in mice and humans and numerous vasculature-targeting peptides have been reported. However, in vivo phage display has not typically produced molecules that extravasate to target specific organ or tumor antigens. Phage selections in animals have been performed for very short times without optimization for biodistribution or clearance rates to a particular organ. It is hypothesized that peptides that home to a desired antigen/organ can be obtained from in vivo phage experiments by optimization of incubation times, phage extraction and propagation procedures. To accomplish this goal, one must first gain a better understanding of the in vivo biodistribution and rate of clearance of engineered phage peptide display libraries. While the fate of wild type phage in rodents has been reported, the in vivo biodistribution of the commonly used engineered fd-tet M13 phage peptide display libraries (such as in the fUSE5 vector system) have not been well established. Here we report the biodistribution and clearance properties of fd-tet fifteen amino acid random peptide display libraries in fUSE5 phage in three common mouse models employed for drug discovery - CF-1, nude, and SCID mice.