cDNA Display Research Articles

C1q potentiates DR5 agonist activity by a mixture of two DR5-specific HexaBody® molecules

Single-domain antibodies (variable domain of the heavy chain of a heavy chain antibody; VHH) are promising reagents for therapeutics and diagnostics because of their stability, cost-effective production and material workability as a small antibody. Currently, general acquisition of a VHH using immunization of camelids is inconvenient from the standpoint of animal protection, cost and the process is time-consuming. Thus, a straightforward and efficient method for screening VHHs against a target molecule is required. In this study, we examined whether in vitro selection of a VHH against a target protein could be performed by a cDNA display method with an artificial VHH library that had the three complementarity-determining regions (CDRs) randomized by chemical synthesis. The results revealed that a particular VHH against survivin, which is a member of the inhibitor of apoptosis family, was selected with affinity in the range of 10−7 to 10−8 M. The in vitro selection of a VHH using cDNA display with an artificial synthesized library without animal immunization was shown to be effective for rapid and inexpensive screening of VHHs against a target protein.

Anti-survivin single-domain antibodies derived from an artificial library including three synthetic random regions by in vitro selection using cDNA display

Single-domain antibodies (variable domain of the heavy chain of a heavy chain antibody; VHH) are promising reagents for therapeutics and diagnostics because of their stability, cost-effective production and material workability as a small antibody. Currently, general acquisition of a VHH using immunization of camelids is inconvenient from the standpoint of animal protection, cost and the process is time-consuming. Thus, a straightforward and efficient method for screening VHHs against a target molecule is required. In this study, we examined whether in vitro selection of a VHH against a target protein could be performed by a cDNA display method with an artificial VHH library that had the three complementarity-determining regions (CDRs) randomized by chemical synthesis. The results revealed that a particular VHH against survivin, which is a member of the inhibitor of apoptosis family, was selected with affinity in the range of 10−7 to 10−8 M. The in vitro selection of a VHH using cDNA display with an artificial synthesized library without animal immunization was shown to be effective for rapid and inexpensive screening of VHHs against a target protein.

Self-assembling functional programmable protein array for studying protein\u2013protein interactions in malaria parasites

BackgroundPlasmodium vivax is the most widespread malarial species, causing significant morbidity worldwide. Knowledge is limited regarding the molecular mechanism of invasion due to the lack of a continuous in vitro culture system for these species. Since protein–protein and host–cell interactions play an essential role in the microorganism’s invasion and replication, elucidating protein function during invasion is critical when developing more effective control methods. Nucleic acid programmable protein array (NAPPA) has thus become a suitable technology for studying protein–protein and host–protein interactions since producing proteins through the in vitro transcription/translation (IVTT) method overcomes most of the drawbacks encountered to date, such as heterologous protein production, stability and purification.ResultsTwenty P. vivax proteins on merozoite surface or in secretory organelles were selected and successfully cloned using gateway technology. Most constructs were displayed in the array expressed in situ, using the IVTT method. The Pv12 protein was used as bait for evaluating array functionality and co-expressed with P. vivax cDNA display in the array. It was found that Pv12 interacted with Pv41 (as previously described), as well as PvMSP142kDa, PvRBP1a, PvMSP8 and PvRAP1.ConclusionsNAPPA is a high-performance technique enabling co-expression of bait and query in situ, thereby enabling interactions to be analysed rapidly and reproducibly. It offers a fresh alternative for studying protein–protein and ligand–receptor interactions regarding a parasite which is difficult to cultivate (i.e. P. vivax).

A System for Global Analysis of Correlation between Protein Expression and mRNA

The influence of mRNA abundance on protein expression continues to be a topic of significant interest and debate. In various systems, the correlations between these molecules has been reported to be anywhere from 20–70%. Many of these studies have been conducted in complex, in vivo systems, and assayed just a fraction of the genes present due to difficulties in quantitation of many proteins simultaneously. Our approach is to examine this relationship as the result of a single translational event, in a global manner, and without the additional factors introduced by in vivo systems. We created a protein library whose constituents can be decoded by next generation sequencing using cDNA display techniques, a synthetic, bacterial based, translation system, and an expression library derived from an mRNA population. This data, along with sequencing data from the precursor mRNA library, allow for a correlation analysis across all genes contained within the mRNA library. Preliminary results indicates that in this system, ~60% of the variance in protein expression can be explained by mRNA abundance. To determine the stability of the correlation, we modified the translation sequences in the gene library and reanalyzed the protein to mRNA relationship; variations in the correlation reflect the influence of the translational sequences on the protein expression. This system can be utilized to explore the mRNA and protein relationship for wild‐type translational systems, or to evaluate the effects of modifications or synthetic sequences.This abstract is from the Experimental Biology 2018 Meeting. There is no full text article associated with this abstract published in The FASEB Journal.

In Vitro Selection of Single-Domain Antibody (VHH) Using cDNA Display.

Single-domain antibody (e.g., Nanobody, VHH antibody) is a promising scaffold for therapeutic and diagnostic reagents. To expand the range of target molecules, in vitro selection using cell-free display technologies such as cDNA display is useful and powerful because of their huge libraries and robust stability. We provide technical details for in vitro selection of single-domain antibodies using cDNA display.

Phage Display cDNA Screens Reveal a Specific Region of Plasmodium falciparum Glutamic Acid‐rich Protein that Binds to Human Red Blood Cells. A Functional Role in Malaria Pathogenesis

Severe malaria caused by Plasmodium falciparum is a serious disease that affects a large number of people regardless of the mode of treatment with currently existing antimalarial drugs. In 2015, the Centers for Disease Control and Prevention (CDC) estimated nearly 214 million cases of malaria infection worldwide resulting in ~400,000 deaths mainly in sub‐Saharan Africa. Infection of red blood cells (RBCs) by P. falciparum, the most lethal species among the human malaria parasites, leads to several complications such as cytoadherence, which involves binding of infected red blood cells (iRBCs) to endothelial receptors, and rosetting, which is the binding of iRBCs to uninfected RBCs. These features contribute to specific organ damage and exacerbate disease pathogenesis. To elucidate the biological mechanisms of P. falciparum infection in human RBCs, we are currently investigating the virulence phenomena including parasite invasion and rosetting. Using phage display technology, we previously identified several novel ligand‐receptor interactions at the blood stage of malaria infection. Here, we used two independent phage display cDNA library screens and identified P. falciparum glutamic acid‐rich protein (Pf‐GARP) as a novel parasite ligand that binds to the surface of human RBCs. Specifically, we screened phage display cDNA libraries constructed from P. falciparum FCR‐3 and 3D7 strains and identified several overlapping Pf‐GARP clones expressing peptides that bind to intact RBCs. Two phage clones encoding Pf‐GARP392–437 and Pf‐GARP356–552 revealed a specific binding region located nearly in the middle of full length Pf‐GARP1–673. Using bioinformatics tools, we chemically synthesized a codon‐optimized segment, Pf‐GARP370–444, located within the overlapping phage clones. The phage cDNA inserts were cloned into pET‐32b vector, and recombinant peptides were expressed in E. coli. Recombinant fusion proteins encoded by all three Pf‐GARP cDNA were confirmed to bind human RBCs in vitro. The codon‐optimized Pf‐GARP370–444 was cloned into the pRE4 vector, and expressed on the surface of Chinese hamster ovary (CHO) cells. The transfected CHO cells formed rosette‐like clusters in the presence of uninfected human RBCs in vitro. Importantly, direct biochemical interaction between purified Pf‐GARP370–444 protein and human RBCs was abolished upon treatment of RBCs with chymotrypsin but not with trypsin and neuraminidase. These results indicate that a centrally‐located segment of Pf‐GARP recognizes a specific chymotrypsin‐sensitive receptor(s) on the surface of host RBCs. Our findings suggest a functional role of Pf‐GARP in the rosetting and invasion phenomena with implications in inter‐membrane transport and vaccine development against malaria.Support or Funding InformationFunding source: NIH‐HL060961

In vitro selection of random peptides against artificial lipid bilayers: a potential tool to immobilize molecules on membranes.

The first in vitro selection of binding peptides against artificial lipid membranes from a random peptide library using an in vitro display method (cDNA display) is reported. The selected peptide, LB-1, has both amphiphilic and cationic regions, and proteins fused to LB-1 can be immobilized on the liposome surface.

Enhanced mRNA-protein fusion efficiency of a single-domain antibody by selection of mRNA display with additional random sequences in the terminal translated regions.

In vitro display technologies such as mRNA and cDNA display are powerful tools to create and select functional peptides. However, in some cases, efficiency of mRNA-protein fusion is very low, which results in decreased library size and lower chance of successful selection. In this study, to improve mRNA-protein fusion efficiency, we prepared an mRNA display library of a protein with random N- and C-terminal coding regions consisting of 12 nucleotides (i.e. four amino acids), and performed an electrophoresis mobility shift assay (EMSA)-based selection of successfully formed mRNA display molecules. A single-domain antibody (Nanobody, or VHH) was used as a model protein, and as a result, a pair of sequences was identified that increased mRNA-protein fusion efficiency of this protein by approximately 20%. Interestingly, enhancement of the fusion efficiency induced by the identified sequences was protein-specific, and different results were obtained for other proteins including VHHs with different CDRs. The results suggested that conformation of mRNA as a whole, rather than the amino acid sequence of the translated peptide, is an important factor to determine mRNA-protein fusion efficiency.

An RNA Binding Peptide Consisting of Four Types of Amino Acid by in Vitro Selection Using cDNA Display.

RNA–protein interactions have a central role in the living world. In this article, we examined whether primitive peptides (30 residues) consisting of four types of amino acid (Gly, Ala, Asp, and Val) could interact with tRNA as a model of primitive RNAs in the RNA world. By in vitro selection of binding peptides using the cDNA display method, a characteristic peptide was selected from a random peptide library and assayed by electrophoretic mobility shift and pull-down assays. Interestingly, the selected peptide bound to a single-stranded region including a loop structure of an RNA molecule with some sequence specificity.

Rapid Discovery of Functional Small Molecule Ligands against Proteomic Targets through Library-Against-Library Screening.

Identifying “druggable” targets and their corresponding therapeutic agents are two fundamental challenges in drug discovery research. The one-bead-one-compound (OBOC) combinatorial library method has been developed to discover peptides or small molecules that bind to a specific target protein or elicit a specific cellular response. The phage display cDNA expression proteome library method has been employed to identify target proteins that interact with specific compounds. Here, we combined these two high-throughput approaches, efficiently interrogated approximately 1013 possible molecular interactions, and identified 91 small molecule compound beads that interacted strongly with the phage library. Of 19 compounds resynthesized, 4 were cytotoxic against cancer cells; one of these compounds was found to interact with EIF5B and inhibit protein translation. As more binding pairs are confirmed and evaluated, the “library-against-library” screening approach and the resulting small molecule–protein domain interaction database may serve as a valuable tool for basic research and drug development.

Proteome-wide Identification of Novel Ceramide-binding Proteins by Yeast Surface cDNA Display and Deep Sequencing

Although the bioactive sphingolipid ceramide is an important cell signaling molecule, relatively few direct ceramide-interacting proteins are known. We used an approach combining yeast surface cDNA display and deep sequencing technology to identify novel proteins binding directly to ceramide. We identified 234 candidate ceramide-binding protein fragments and validated binding for 20. Most (17) bound selectively to ceramide, although a few (3) bound to other lipids as well. Several novel ceramide-binding domains were discovered, including the EF-hand calcium-binding motif, the heat shock chaperonin-binding motif STI1, the SCP2 sterol-binding domain, and the tetratricopeptide repeat region motif. Interestingly, four of the verified ceramide-binding proteins (HPCA, HPCAL1, NCS1, and VSNL1) and an additional three candidate ceramide-binding proteins (NCALD, HPCAL4, and KCNIP3) belong to the neuronal calcium sensor family of EF hand-containing proteins. We used mutagenesis to map the ceramide-binding site in HPCA and to create a mutant HPCA that does not bind to ceramide. We demonstrated selective binding to ceramide by mammalian cell-produced wild type but not mutant HPCA. Intriguingly, we also identified a fragment from prostaglandin D2synthase that binds preferentially to ceramide 1-phosphate. The wide variety of proteins and domains capable of binding to ceramide suggests that many of the signaling functions of ceramide may be regulated by direct binding to these proteins. Based on the deep sequencing data, we estimate that our yeast surface cDNA display library covers ∼60% of the human proteome and our selection/deep sequencing protocol can identify target-interacting protein fragments that are present at extremely low frequency in the starting library. Thus, the yeast surface cDNA display/deep sequencing approach is a rapid, comprehensive, and flexible method for the analysis of protein-ligand interactions, particularly for the study of non-protein ligands.

Profiling lethal factor interacting proteins from human stomach using T7 phage display screening.

The anthrax lethal factor (LF) is a zinc dependent metalloproteinase that cleaves the majority of mitogen-activated protein kinase kinases and a member of NOD-like receptor proteins, inducing cell apoptosis. Despite efforts to fully understand the Bacillus anthracis toxin components, the gastrointestinal (GI) anthrax mechanisms have not been fully elucidated. Previous studies demonstrated gastric ulceration, and a substantial bacterial growth rate in Peyer's patches. However, the complete molecular pathways of the disease that results in tissue damage by LF proteolytic activity remains unclear. In the present study, to identify the profile of the proteins potentially involved in GI anthrax, protein-protein interactions were investigated using human stomach T7 phage display (T7PD) cDNA libraries. T7PD is a high throughput technique that allows the expression of cloned DNA sequences as peptides on the phage surface, enabling the selection and identification of protein ligands. A wild type and mutant LF (E687A) were used to differentiate interaction sites. A total of 124 clones were identified from 194 interacting-phages, at both the DNA and protein level, by in silico analysis. Databases revealed that the selected candidates were proteins from different families including lipase, peptidase-A1 and cation transport families, among others. Furthermore, individual T7PD candidates were tested against LF in order to detect their specificity to the target molecule, resulting in 10 LF-interacting peptides. With a minimum concentration of LF for interaction at 1 μg/ml, the T7PD isolated pepsin A3 pre-protein (PAP) demonstrated affinity to both types of LF. In addition, PAP was isolated in various lengths for the same protein, exhibiting common regions following PRALINE alignment. These findings will help elucidate and improve the understanding of the molecular pathogenesis of GI anthrax, and aid in the development of potential therapeutic agents.

A High Performance Platform Based on cDNA Display for Efficient Synthesis of Protein Fusions and Accelerated Directed Evolution.

We describe a high performance platform based on cDNA display technology by developing a new modified puromycin linker-oligonucleotide. The linker consists of four major characteristics: a "ligation site" for hybridization and ligation of mRNA by T4 RNA ligase, a "puromycin arm" for covalent linkage of the protein, a "polyadenosine site" for a longer puromycin arm and purification of protein fusions (optional) using oligo-dT matrices, and a "reverse transcription site" for the formation of stable cDNA protein fusions whose cDNA is covalently linked to its encoded protein. The linker was synthesized by a novel branching strategy and provided >8-fold higher yield than previous linkers. This linker enables rapid and highly efficient ligation of mRNA (>90%) and synthesis of protein fusions (∼ 50-95%) in various cell-free expression systems. Overall, this new cDNA display method provides 10-200 fold higher end-usage fusions than previous methods and benefits higher diversity libraries crucial for directed protein/peptide evolution. With the increased efficiency, this system was able to reduce the time for one selection cycle to <8 h and is potentially amenable to high-throughput systems. We demonstrate the efficiency of this system for higher throughput selections of various biomolecular interactions and achieved 30-40-fold enrichment per selection cycle. Furthermore, a 4-fold higher enrichment of Flag-tag was obtained from a doped mixture compared with that of the previous cDNA display method. A three-finger protein library was evolved to isolate superior nanomolar range binding candidates for vascular endothelial growth factor. This method is expected to provide a beneficial impact to accelerated drug discovery and proteome analysis.

PEP-on-DEP: A competitive peptide-based disposable electrochemical aptasensor for renin diagnostics

Antibody-based immunosensors are relatively less accessible to a wide variety of unreachable targets, such as low-molecular-weight biomarkers that represent a rich untapped source of disease-specific diagnostic information. Here, we present a peptide aptamer-based electrochemical sensor technology called ‘PEP-on-DEP’ to detect less accessible target molecules, such as renin, and to improve the quality of life. Peptide-based aptamers represent a relatively smart class of affinity binders and show great promise in biosensor development. Renin is involved in the regulation of arterial blood pressure and is an emerging biomarker protein for predicting cardiovascular risk and prognosis. To our knowledge, no studies have described aptamer molecules that can be used as new potent probes for renin. Here, we describe a portable electrochemical biosensor platform based on the newly identified peptide aptamer molecules for renin. We constructed a randomized octapeptide library pool with diversified sequences and selected renin specific peptide aptamers using cDNA display technology. We identified a few peptide aptamer sequences with a KD in the µM binding affinity range for renin. Next, we grafted the selected peptide aptamers onto gold nanoparticles and detected renin in a one-step competitive assay using our originally developed DEP (Disposable Electrochemical Printed) chip and a USB powered portable potentiostat system. We successfully detected renin in as little as 300ngmL−1 using the PEP-on-DEP method. Thus, the generation and characterization of novel probes for unreachable target molecules by merging a newly identified peptide aptamer with electrochemical transduction allowed for the development of a more practical biosensor that, in principle, can be adapted to develop a portable, low-cost and mass-producible biosensor for point-of-care applications.

A versatile puromycin-linker using cnvK for high-throughput in vitro selection by cDNA display

cDNA display is a powerful in vitro display technology used to explore functional peptides and proteins from a huge library by in vitro selection. In addition to expediting the in vitro selection cycle by using cDNA display, easy and rapid functional analysis of selected candidate clones is crucial for high-throughput screening of functional peptides and proteins. In this report, a versatile puromycin-linker employing an ultrafast photo-cross-linker, 3-cyanovinylcarbazole nucleoside, is introduced. Its utility for both in vitro selection using cDNA display and protein–protein interaction analysis using a surface plasmon resonance (SPR) system is described. Using this versatile puromycin-linker, we demonstrated the model in vitro selection of the FLAG epitope and a SPR-based assay to measure the dissociation constant between the B domain of protein A and immunoglobulin G. Improvement of the puromycin-linker as described herein should make the cDNA display method easier to utilize for design of protein or peptide based affinity reagents.

Suppression of intestinal immunity through silencing of TCTP by RNAi in transgenic silkworm, Bombyx mori

Intestinal immune response is a front line of host defense. The host factors that participate in intestinal immunity response remain largely unknown. We recently reported that Translationally Controlled Tumor Protein (BmTCTP) was obtained by constructing a phage display cDNA library of the silkworm midgut and carrying out high throughput screening of pathogen binding molecules. To further address the function of BmTCTP in silkworm intestinal immunity, transgenic RNAi silkworms were constructed by microinjection piggBac plasmid to Dazao embryos. The antimicrobial capacity of transgenic silkworm decreased since the expression of gut antimicrobial peptide from transgenic silkworm was not sufficiently induced during oral microbial challenge. Moreover, dynamic ERK phosphorylation from transgenic silkworm midgut was disrupted. Taken together, the innate immunity of intestinal was suppressed through disruption of dynamic ERK phosphorylation after oral microbial infection as a result of RNAi-mediated knockdown of midgut TCTP in transgenic silkworm.

Complementary DNA display selection of high-affinity peptides binding the vacuolating toxin (VacA) of Helicobacter pylori.

Artificial peptides designed for molecular recognition of a bacterial toxin have been developed. Vacuolating cytotoxin A protein (VacA) is a major virulence factor of Helicobacter pylori, a gram-negative microaerophilic bacterium inhabiting the upper gastrointestinal tract, particularly the stomach. This study attempted to identify specific peptide sequences with high affinity for VacA using systematic directed evolution in vitro, a cDNA display method. A surface plasmon resonance-based biosensor and fluorescence correlation spectroscopy to examine binding of peptides with VacA identified a peptide (GRVNQRL) with high affinity. Cyclization of the peptide by attaching cysteine residues to both termini improved its binding affinity to VacA, with a dissociation constant (Kd ) of 58 nm. This study describes a new strategy for the development of artificial functional peptides, which are promising materials in biochemical analyses and medical applications.

A novel zinc finger gene, ZNF465, is inappropriately expressed in acute myeloid leukaemia cells.

To increase our knowledge of leukaemia-associated antigens, especially in acute myeloid leukaemia (AML) M4, we prepared a phage display cDNA library using mRNA from the bone marrow cells of a patient with AML M4 at diagnosis. We immunoscreened 10(6) pfu with autologous sera and identified an antigen which we named GKT-AML8. The cDNA showed more than 99% similarity to a sequence on 2q21.2 and 95% sequence similarity to a sequence on 19q13.3. These genes were named ZNF465 and ZNF466, respectively, following HUGO Gene Nomenclature Committee (HGNC) guidelines. Expressed sequence tag data suggests that both genes are transcriptionally active. ZNF465 and ZNF466 encode a 5' krüppel associated box domain typical of negative regulators of gene transcription. We have confirmed the translational start site in the +1 frame in a near-Kozak sequence that produces a 102 amino acid polypeptide from ZNF465. The high level of sequence similarity between ZNF465 and ZNF466 makes their transcripts almost indistinguishable by real-time polymerase chain reaction (RT-PCR). However, GKT-AML8 showed most sequence similarity to ZNF465 and no transcript matching the 3' ZNF466 sequence could be detected in patient samples or healthy volunteers. ZNF465/466 expression was detectable in 12/13 AML and 10/14 chronic myeloid leukaemia patients' samples but not in normal donor peripheral blood (0/8) or 0/3 bone marrow samples which had been separated into CD34(+) and CD34(-) samples. The altered expression of ZNF465/466 in patients' samples and its absence in healthy donor haematopoietic samples indicate that ZNF465 is overexpressed in early myeloid disease and as such may represent a promising target for immunotherapy.

A 24-48 h fed Amblyomma americanum tick saliva immuno-proteome.

BackgroundMultiple tick saliva proteins, the majority of which are unknown, confer tick resistance in repeatedly infested animals. The objective of this study was to identify the 24-48 h fed Amblyomma americanum tick saliva immuno-proteome. The 24-48 h tick-feeding phase is critical to tick parasitism as it precedes important events in tick biology, blood meal feeding and disease agent transmission. Fed male, 24 and 96 h fed female phage display cDNA expression libraries were biopanned using rabbit antibodies to 24 and 48 h fed A. americanum female tick saliva proteins. Biopanned immuno-cDNA libraries were subjected to next generation sequencing, de novo assembly, and bioinformatic analysis.ResultsMore than 800 transcripts that code for 24-48 h fed A. americanum immuno-proteins are described. Of the 895 immuno-proteins, 52% (464/895) were provisionally identified based on matches in GenBank. Of these, ~19% (86/464) show high level of identity to other tick hypothetical proteins, and the rest include putative proteases (serine, cysteine, leukotriene A-4 hydrolase, carboxypeptidases, and metalloproteases), protease inhibitors (serine and cysteine protease inhibitors, tick carboxypeptidase inhibitor), and transporters and/or ligand binding proteins (histamine binding/lipocalin, fatty acid binding, calreticulin, hemelipoprotein, IgG binding protein, ferritin, insulin-like growth factor binding proteins, and evasin). Others include enzymes (glutathione transferase, cytochrome oxidase, protein disulfide isomerase), ribosomal proteins, and those of miscellaneous functions (histamine release factor, selenoproteins, tetraspanin, defensin, heat shock proteins).ConclusionsData here demonstrate that A. americanum secretes a complex cocktail of immunogenic tick saliva proteins during the first 24-48 h of feeding. Of significance, previously validated immunogenic tick saliva proteins including AV422 protein, calreticulin, histamine release factor, histamine binding/lipocalins, selenoproteins, and paramyosin were identified in this screen, supporting the specificity of the approach in this study. While descriptive, this study opens opportunities for in-depth tick feeding physiology studies.Electronic supplementary materialThe online version of this article (doi:10.1186/1471-2164-15-518) contains supplementary material, which is available to authorized users.

Template-directed chemical ligation to obtain 3'-3' and 5'-5' phosphodiester DNA linkages.

Up to now, the direct ligation of two DNA fragments with opposite directions to obtain 3′-3′ or 5′-5′ phosphate ester bonds is still challenging. The only way to obtain DNA oligonucleotides containing a 3′-3′ or 5′-5′ inversion of polarity sites is based on professional DNA chemical synthesis. Herein, we demonstrate a convenient template-directed chemical ligation that enables 3′-3′ and 5′-5′ linkages of two DNA oligonucleotides. This method is based on the assembly of two oligonucleotides on a template in opposite directions through forming antiparallel and parallel duplexes simultaneously, followed by coupling with N-Cyanoimidazole under mild condition. Moreover, on the basis of DNA oligonucleotides with 5′-5′ linkage obtained through our template-directed chemical ligation, we developed a new cDNA display technique for in vitro selection of functional polypeptides.