Development of Anti-Infectives Using Phage Display: Biological Agents against Bacteria, Viruses, and Parasites

Abstract

The vast majority of anti-infective therapeutics on the market or in development are small molecules; however, there is now a nascent pipeline of biological agents in development. Until recently, phage display technologies were used mainly to produce monoclonal antibodies (MAbs) targeted against cancer or inflammatory disease targets. Patent disputes impeded broad use of these methods and contributed to the dearth of candidates in the clinic during the 1990s. Today, however, phage display is recognized as a powerful tool for selecting novel peptides and antibodies that can bind to a wide range of antigens, ranging from whole cells to proteins and lipid targets. In this review, we highlight research that exploits phage display technology as a means of discovering novel therapeutics against infectious diseases, with a focus on antimicrobial peptides and antibodies in clinical or preclinical development. We discuss the different strategies and methods used to derive, select, and develop anti-infectives from phage display libraries and then highlight case studies of drug candidates in the process of development and commercialization. Advances in screening, manufacturing, and humanization technologies now mean that phage display can make a significant contribution in the fight against clinically important pathogens.

INTRODUCTION

Infectious diseases continue to be one of the leading causes of human mortality and disability worldwide despite the increasing availability of vaccines. In today's interconnected world, infectious diseases are able to spread rapidly and globally and also appear to be emerging more frequently (50). For example, new infectious diseases have been identified at the rate of more than one per year from the 1970s to the 1990s (120), and more have recently emerged, with some lethal ones, such as severe acute respiratory syndrome (SARS) and avian influenza, triggering major international concern (74, 77, 116). Moreover, in 2001, the anthrax letter incidents highlighted the potential threat posed by the malicious release of biological threat agents (8, 68). Today, gains being made in many areas of infectious disease control are also being seriously jeopardized by the spread of antimicrobial resistance, with drug resistance now being a concern for many pathogens, including methicillin-resistant Staphylococcus aureus (MRSA), vancomycin-resistant Enterococcus faecium (VRE), carbapenem-resistant Klebsiella pneumoniae (NDM-1), and multidrug-resistant (MDR) Pseudomonas aeruginosa. A public health crisis is looming, as this development of “superbugs” is coupled with a weak pipeline of new antimicrobials reaching the marketplace, leading to the World Health Organization declaring antimicrobial resistance to be one of the greatest threats to human health (153).

The first bacterial genome, sequenced in 1995, led to the prospect of hundreds of new genes to explore as targets, with the potential advantage of being able to directly screen for therapeutic candidates against targets identified from the genome. However, despite the wealth of genomic data available for many pathogens, target-based screening has been significantly less successful than expected in finding novel acting anti-infectives (100, 107). Instead, most of the antimicrobials reaching the marketplace in recent years have been derivative or expanded-spectrum drugs, with very few novel anti-infectives being discovered (15, 16, 27, 28). The shortage of new antibiotics is also exacerbated by the exit of many pharmaceutical companies from the field despite an unmet need for such compounds (7).

Broad-spectrum antibiotics have, until recently, been the mainstay of antibiotic therapy. Their commercial success has limited the interest in developing pathogen-specific drugs, as exemplified by pathogen-specific antibodies and other biopharmaceuticals. However, the increasing threat of drug-resistant bacteria has forced a rethink on the best strategies for treating them, increasing the awareness of pathogen-specific therapies. Such strategies include the screening of large combinatorial chemical libraries, natural products, and biologics for their ability to inhibit the growth of specific pathogens in whole-cell assays. Compared to small-molecule anti-infective discovery, biopharmaceuticals still occupy a relatively new clinical development and commercial niche. The emergence of new recombinant technologies, such as phage display, for the discovery of monoclonal antibodies (MAbs) and peptides has provided significant opportunities for the development of more specific antimicrobials.

Since its invention in 1985 (128), phage display has been successfully applied to many different areas of research, including immunology, cancer research, drug discovery, epitope mapping, protein-protein interactions, plant sciences, and infectious diseases, targeting a broad cross-section of protein families. It has also been used to identify small peptide ligands and antibodies inhibiting the function of targeted receptors for a wide range of applications. Since the 1990s, phage display has shown great potential in the discovery of new therapeutics (26, 42). Some peptides and MAbs derived from phage display are currently in clinical or preclinical trials (42, 113). Phage display is now playing a significant role for the discovery of peptides and antibodies that may serve as novel therapeutics (37, 91). In this review, we will summarize phage display literature focused on antimicrobial studies during the past 2 decades and review the application of the technology in industry for developing new anti-infective drugs.

PHAGE AND PHAGE DISPLAY

M13 phage.

Phage display technology is based on the construction of a polypeptide library fused to a bacteriophage coat protein. Although T4, T7, and λ phage have been used for phage display (35, 133, 135), the most commonly used phages are M13 and fd filamentous phage because they do not lyse infected bacteria during their life cycle (Fig. 1). Filamentous bacteriophages consist of a genome of circular single-stranded DNA (ssDNA) and a flexible rod-shaped cylinder approximately 1 μm long and 6 nm in diameter which is composed of five coat proteins. All the coat proteins can be used for display, but pVIII and pIII are the most commonly applied. Each M13 virion contains 2,700 copies of pVIII protein which together compose 87% of the total virion mass. The pVIII protein is largely α-helical and rod shaped, with approximately 50% of its 50 amino acids being surface exposed, making it suitable for display. However, pVIII is limited to displaying short peptide sequences due to virion package reasons. On the other hand, pIII is an ideal option for displaying large insertions, yet the infectivity of phages may be reduced.

Fig 1.

Life cycle of filamentous phages. Filamentous phage binds to the F pilus of a host E. coli cell through pIII. Then the host TolA protein starts to depolymerize the phage coat proteins, which remain in the inner membrane for recycling. The ssDNA of the phage enters into the cytoplasm, converts into double-stranded DNA (dsDNA), and starts replication and expression using host enzymes. ssDNA and coated pV protein dimers form the precursors of the phage. Then pV is replaced by pVIII in the channel formed by pI, pXI, pIV, and host thioredoxin; in the meantime, mature phage particles are assembled and released.

Library construction.

The general procedures of the phage display experiment consist of three stages: (i) construction of a library with peptide or antibody variants, (ii) selections based on affinity to interested targets, and (iii) confirmation of selected binders using biological assays and analysis. For the construction of a library, it is important to first consider which system is most suitable for the desired end product. There are three general classes of phage display systems. The first is based on the natural filamentous phage genome, the ssDNA vector. Libraries constructed by introducing foreign DNA inserts into the phage genome will result in the fusion gene product displayed on all the coat proteins. The second system entails the use of plasmid vectors, also known as phagemids. A phagemid generally contains bacterial and phage origins of replication, an antibiotic resistance gene, and the fusion gene with a weak promoter. Third, a “hybrid system,” which still utilizes the phage genome but which contains both a wide-type phage gene and a fusion gene, can be employed (167). To distinguish between these systems, Smith coined the terms “3,” “3 + 3,” and “33,” respectively (129) (Fig. 2). Numbers indicate the coat protein. For example, if the library is constructed on pVIII, the formats are “8,” “8 + 8,” and “88.” In general, fusion library DNA on phage vectors with natural phage promoters will produce a polyvalent display on the phage surface, whereas the phagemid vectors and hybrid phage vectors always lead to a monovalent display. In addition, because a phagemid vector contains only a fusion gene, it needs a helper phage, which is a filamentous phage with reduced packaging efficiency, to encapsidate into phage particles. The valency of the display links directly to the affinity of the binders. Monovalent display systems are more suitable for the identification of the strongest binders because they allow for selection based on pure affinity, whereas polyvalent display prevents the highest-affinity clones in a selection from being identified because it confers a high apparent affinity on weak-binding clones. Rondot and colleagues developed Hyperphage, which allows the use of monovalent libraries to select high-avidity binders (115).

Fig 2.

Library construction systems. Black boxes indicate the gene fragments encoding pIII. Yellow boxes represent the foreign gene inserted into the pIII gene. Yellow circles show the fusion proteins expressed on the N terminus of the pIII protein.

It is feasible to display peptides and functional antibodies with a wide range of sizes and structures on a phage surface. However, practically, not all peptides or proteins can be selected using phage display. Some sequences may be toxic to the expression vector (normally Escherichia coli), interfere with phage assembly, or be sensitive to bacterial proteases. In addition, cysteine residues are often rarer than expected based on predictions of randomness in peptide libraries, a result which has been attributed to reduced display due to the formation of inappropriate disulfide bonds.

Similar to other selection technologies, the size and quality of libraries are crucial for the success of phage display. Random peptide libraries are the most common type of phage display libraries. Currently, methods for the production of peptide libraries are still based on the procedure developed by Zoller and Smith (170, 171), subsequently improved by Kunkel (63). They employed artificially synthesized oligonucleotides to introduce mutations into the vectors, a process which can lead to library sizes from 10⁹ to 10¹¹; however, some libraries may contain up to 20% of wild-type phage. By adding an amber stop codon (TAG) in the beginning of gene III of M13 phage, Scholle and colleagues generated peptide libraries with 100% recombinant phages (124). In 1990, McCafferty and colleagues successfully displayed antibody variable domains on the surface of filamentous phage (83). Since then, single-chain variable fragments (scFvs) or antigen-binding fragments (Fabs) have generally been used for the construction of an antibody phage display library, because the complementarity-determining regions (CDRs) located at the variable domains of both heavy chain (VH) and light chain (VL) determine the binding of antigens. The construction of antibody libraries has been extensively reviewed (94, 103, 146). The traditional approach, based on the method established by Marks and coworkers in 1991 (76), uses PCR assembly to randomly combine VH and VL genes into phagemid vectors that produce human scFv repertoires. Libraries generated using the traditional method generally contain 10⁷ members, a size which is not sufficient enough for screening. Optimized and novel strategies to generate large antibody libraries have been carried out, increasing the diversity to 10¹² (23, 39, 121, 126). However, both monovalently and polyvalently displayed scFvs and Fabs showed limitations that make them difficult for the selections based on intrinsic affinity (103). Furthermore, the isolated antibody fragments require additional cloning steps and reformatting in mammalian cells to generate full-length IgGs which are suitable for therapeutic applications. Thus, libraries with functional full-length antibodies displayed on phage particles may become much more ideal as screening tools. Mazor and colleagues developed an E. coli-based protein expression technology, named E-clonal, which enables the isolation of full-length IgGs from libraries displayed in the periplasm of E. coli cells through fluorescence-activated cell sorting (FACS) (80, 81). In the E-clonal system, a phagemid, pMAZ360-IgG, was constructed for efficient expression of soluble intact IgGs in E. coli, and then antigen-specific IgGs were isolated by the binding to fluorescently labeled antigen. However, FACS is not efficient enough to sort a library with more than 10⁹ cells. In order to solve the problem, a prescreening step of using phage display panning to reduce the library size was carried out (79). A fUSE5-ZZ phage was introduced into the system, making it possible to display full-length IgGs on phage particles. The fUSE5-ZZ phage particles display the fragment crystallizable region (Fc)-binding ZZ protein on the pIII protein. By the infection of fUSE5-ZZ phage, soluble IgGs expressed in pMAZ360-IgG-transformed E. coli cells were captured on the phage particles, leading to a full-length IgG phage display library. Combined with subsequent FACS screening, low-nanomolar-range antibodies were isolated from the library (79). The system flexibly applies fUSE5-ZZ phage as both a helper and a capture phage. It significantly contributes to the development of full-length antibody phage display libraries.

At present, although hundreds of libraries were constructed in different research groups, the resources of commercial libraries are very limited. The most commonly used commercial libraries are New England BioLabs (NEB) Ph.D. phage display libraries, which have a diversity of 10⁹ independent peptides.

Biopanning.

The phage selection procedure, also referred as biopanning, is a highly flexible and dynamic step in a phage display experiment. It is based on affinity selection, which is a characteristic aspect of the phage display technology that selects for ligands against any target. By permitting control over selection and screening conditions, display technologies allow the generation of antibodies/peptides against defined antigen conformations or epitopes. In vitro methods also overcome immunological tolerance, allowing the selection of affinity reagents that recognize highly conserved targets (9).

A selection cycle basically contains four stages: (i) incubation of target molecules with a phage display library, (ii) washing off unbound phage, (iii) elution of the bound phage, and (iv) amplification of the eluted phage. Three to four rounds of selections are generally carried out in a biopanning experiment (Fig. 3). In addition, counterscreening selections against carrier molecules or nonspecific targets can also be included to increase the efficiency of biopanning. Phage particles whose displayed peptides/antibodies bind the selector are captured, while all other phages are washed away. The captured phage, generally a 10⁻⁸ to 10⁻⁷ fraction of the initial library population, can then be eluted with mild acid, alkaline, or detergent solutions without affecting phage infectivity.

Fig 3.

Biopanning of a phage display library to select phage binding to an immobilized target.

Although the principle of biopanning is simple, the outcome can vary due to multiple factors, such as library complexity, nature of the target, binding affinity and avidity, and other multiple experimental parameters. Even with an excellent selection strategy, the experiment will fail if the desired peptide/antibody is not present in the library. Binding affinity and avidity are other factors that need to be taken into consideration. As described above, the valency of a displayed molecule on a virion is a critical parameter in the ranking and selection of candidates. High-affinity binders are usually selected from a “3 + 3” system, while multivalent libraries are generally used when seeking high-avidity binders. In addition, phage libraries usually contain a population of “target-unrelated phages” (TUPs), which can bind tightly to other materials used in the panning procedure, such as plastic plates, streptavidin, protein A, bovine serum albumin (BSA), etc. This can be avoided by additional negative screening steps before the biopanning of desired targets. Typical TUPs found in screening campaigns from phage display libraries have been summarized by Menendez and Scott (84).

Most protein, peptide, nucleic acid, and carbohydrate targets can be used with phage display screening. For soluble targets, there are many options for their immobilization. For example, they may be coated directly on a solid surface, such as a polystyrene tube or microplate, or they may be biotinylated and captured on a streptavidin-coated plate or streptavidin-agarose beads (125). Direct coating is easy to perform but may result in an inaccessible ligand binding site, while the latter method involves additional steps for the chemical modification of targets. In the case of insoluble materials, biopanning can be performed in a suspension. In most cases, selection and screening are performed on a solid surface, so it is important to use appropriate blocking methods to avoid the enrichment of target-unrelated peptides or antibodies (84). Historically, these methods have been important for isolating peptides and antibodies against known targets, such as proteins, lipids, and small molecules. While standard phage display methods use purified antigens, panning against whole cells is another option (149). In the past decade, most researchers used whole bacterial/mammalian cells as targets to identify receptor-specific or cell-type-specific peptides and antibodies (5, 18, 102, 134).

The advantages of whole-cell phage display are easy to see. In the case of when antigen is unavailable or the antigen is not stable under immobilization conditions, whole-cell phage display panning is normally the best choice. In addition, it is also useful in the discovery of unknown antigens. The biopanning procedure typically requires no prior knowledge of the cell surface biomarkers, allowing for the isolation of targeting peptides for cell types for which little is known about the cellular profile. For whole-cell screening, the cellular targets are identified in a two-part process. First, peptides or antibodies are first identified by screening whole cells against a phage display library. Second, the binding peptides or antibodies are tested individually in functionally based screens. In all cases, activity is confirmed in functional assays; one does not need to either purify or identify a particular receptor in advance. Since whole cells are used as the affinity matrix, the receptors are likely to be in their native conformation, and a large variety of receptors are being screened at one time. It should be noticed that the cell surfaces would share a high degree of similarity and that the peptides would be recognizing abundant, common receptors. Thus, additional negative selections are necessary to avoid unexpected cell specificity of selected peptides. Nevertheless, it has been reported that the specificity of phage selected in nonbiased screens is 10- to 100-fold higher for the targeted cell type than for other cell types (11). As such, cell-specific peptides are often more useful for targeting certain cell types and for delivery of drugs. With the advent of in vivo panning, it is also possible to target specific cell types or even organs. For example, Apar and colleagues used in vivo screening to identify tumor vasculature-targeted peptides, a process which significantly increased the efficiency of an antitumor drug (3). Four years later, they successfully separated organ-specific peptides by injecting a phage library into the vasculature of a human volunteer (2).

PATHOGEN-TARGETED PHAGE DISPLAY

Phage display has been used widely for identification of specific peptides and antibodies against pathogen targets. These targets are generally subdivided into two categories: (i) molecular targets, such as replication/cell division enzymes and host-pathogen virulence factors, and (ii) whole bacterial cells (Table 1). In comparison with specific molecular targets, cell-based screening has the advantage in that it is an assumption-free strategy with the potential to recognize cell surface structures that may not have been considered targets using genomic-based approaches or that have not yet been identified. Using live pathogens as the target also has the advantage that all “druggable” targets on the cell surface are screened simultaneously in their native physiological context, thus allowing for the selection of potential antimicrobial activity from the outset. Antigens on the cell surface of pathogens are appealing targets for biologics because they provide potential binding sites for molecules to interfere with bacterial division (71), colonization, and virulence (111). Both strategies have been widely applied for developing novel diagnostic tools and therapeutic treatments for infectious diseases. The following section will review key achievements in this area in recent years.

Table 1.

Phage display targets of infectious diseases

Targeta	Libraryb	Potential application	Reference(s)
Molecular targets
TEM-1 β-lactamase	BLIP library	Novel antimicrobial candidates	55, 163
RAP	12-mer peptide library	Anti-S. aureus	162
SEB	12-mer peptide library	Anti-S. aureus	132
S. aureus SdrC	12-mer peptide library	Anti-S. aureus	4
P. aeruginosa MurA	C7C cyclic peptide, 12-mer peptide library	Anti-P. aeruginosa	87
P. aeruginosa MurC	C7C cyclic peptide, 12-mer peptide library	Anti-P. aeruginosa	36
P. aeruginosa MurE and MurF	12-mer peptide library	Anti-P. aeruginosa	96, 97
P. aeruginosa FtsA and FtsZ	C7C cyclic peptide, 12-mer peptide library	Anti-P. aeruginosa	98, 99
H. pylori urease holoenzyme	25-mer peptide, 6-mer peptide library, scFv library	Anti-H. pylori	53, 54, 112
H. pylori surface protein	scFv library	Anti-H. pylori	17
Bacterial membrane model	T7 phage system, 12-mer peptide library	Antibiotic design	136, 158
LPS/lipid A	scFv libraries, peptide libraries	Anti-Gram-negative bacterial agents	43, 44, 48, 59, 60, 78, 85, 92, 137, 169
HCV proteins	scFv libraries, peptide libraries	Epitope mapping, diagnostics, vaccine design	12, 13, 101, 165
HBsAg	scFv library	Drug delivery	152
HBcAg	C7C cyclic peptide library	Anti-HBV infection	51
HAV antibodies	9-mer peptide library	Epitope mapping, diagnostics	66
HEV capsid protein	Antibody library	Anti-HEV infection	123
H5N1 HA	Fab library	Anti-H5N1 infection	69
H5N1 antibodies, anti-H5N1 human sera	H5N1 whole-genome-fragment libraries	Epitope mapping, diagnostics, vaccine design	58
HIV-1 RT	Peptide library	Epitope mapping, vaccine design	34
HIV-1 V3 loop peptide	scFv library	Anti-HIV infection, vaccine design	138
HIV-1 vif protein	12-mer peptide library	Anti-HIV infection	161
HIV-1 IZN17 peptide	Cyclic 10-mer/8-mer peptide library	Anti-HIV infection	151
PRRSV N protein	12-mer peptide library	Detection of PRRSV	114
Malaria peptide mimics/Plasmodium falciparum proteins	scFv libraries	Epitope mapping, therapeutic agents	22, 29, 143
Cryptosporidium parvum glycoproteins	Fab library	Treatment and diagnostics of cryptosporidiosis	24
Whole-cell targets
S. aureus	12-mer peptide library	Anti-S. aureus drug design	159, 160
L. monocytogenes	scFv library	Diagnostics of L. monocytogenes infection	90, 95
L. monocytogenes	12-mer peptide library	Discovery of novel druggable targets	38
L. monocytogenes	9-mer peptide, 12-mer peptide library	Diagnostics of L. monocytogenes infection	21
P. aeruginosa	9-mer peptide, 12-mer peptide library	Diagnostics of P. aeruginosa infection	20
H. pylori	scFv library	Diagnostics of H. pylori infection	118
P. aeruginosa, Enterobacteriaceae	10-mer peptide library	Anti-Gram-negative bacterial agents	104
S. Typhimurium	8-mer peptide library	Diagnostics of S. Typhimurium	131
M. arginini	20-mer peptide library	Drug delivery	33
H. influenzae	15-mer peptide library	Anti-H. influenzae agents	6
C. jejuni	15-mer peptide library	Anti-C. jejuni agents	5
Spores of Bacillus	scFv libraries, peptide libraries	Detection of Bacillus species	10, 62, 140, 144, 154, 168
hCD81-transfected NIH/3T3 cell line	9-mer peptide library	Anti-HCV infection	18
H9N2 virus particles	7-mer peptide library	Anti-H9N2 infection	108, 109
VEEV	scFv library	Detection of VEEV infection	61
C. trachomatis EBs	scFv library	Detection of C. trachomatis	70
Rabies virus	scFv library	Rabies postexposure prophylaxis	166
Andes virus	Cyclic C7C peptide library	Therapeutic agents	45, 46
Sporozoites of E. acervulina	7-mer/C7C/12-mer peptide library	Therapeutic agents	32
Sporozoites of C. parvum	Fab library	Treatment and diagnostics of cryptosporidiosis	24
Mosquito salivary glands	12-mer peptide library	Therapeutic agents	41

^aRAP, RNAIII-activating protein; SEB, staphylococcal enterotoxin B; RT, reverse transcriptase; vif, virion infectivity factor.

^bBLIP, β-lactamase inhibitory protein.

Gram-positive bacteria.

Staphylococcus aureus, one of the most important human pathogens, has become a major threat to human health, as a large percentage of S. aureus infections are mediated by MRSA. In addition, strains resistant to vancomycin, considered one of the “last-resort” antibiotics, have been reported since the 1980s (150). Yacoby and colleagues used S. aureus-specific phage as a carrier to deliver high concentrations of antibiotics to bacterial cells (160). They panned a 12-mer peptide phage display library to isolate the S. aureus-specific phages and also expressed ZZ domains, which bind tightly to immunoglobulins, on the N terminus of the M13 major coat protein pIII. The drug was conjugated to the N terminus of coat protein pVIII using an ester bond. The carrying capacity of the phage was improved by linking the phage to the aminoglycoside antibiotic neomycin (159). The drug molecule is released from the phage surface by ester bond hydrolysis in the presence of serum. Inhibition of bacterial growth was significantly improved by incubating with drug-carrying phages instead of free drug alone. Other studies have employed phage display technology to identify small peptides that bind to S. aureus, S. aureus membrane proteins, or S. aureus toxins (4, 55, 132, 162, 163). Those peptides showed an inhibitory effect on S. aureus growth. However, little research has been done on MRSA or other resistant strains to identify strain-specific peptides or antibodies.

Listeria monocytogenes is a severe food-borne pathogen that causes life-threatening listeriosis. To avoid infection by L. monocytogenes, it is important to detect low levels of the pathogen in food samples. Paoli and colleagues used phage display to identify a scFv antibody that can bind to L. monocytogenes but not other Listeria strains (95). Three years later, a surface plasmon resonance (SPR) sensor was developed based on the scFv antibody (90). L. monocytogenes-specific scFv-displayed phage was immobilized on the sensor surface to detect L. monocytogenes at a detection limit of 2 × 10⁶ CFU/ml. The mechanism of binding to and invading host cells by L. monocytogenes is not clear, limiting the development of appropriate preventative and therapeutic strategies. Exploiting the NEB Ph.D.-12 phage-displayed peptide library, Gasanov and colleagues revealed that invasion by L. monocytogenes is mediated via binding to the insulin-like growth factor II receptor (IGFIIR) on mammalian cells (38). Additional peptides that bound to L. monocytogenes were identified by panning a 9-mer peptide and 12-mer peptide phage display library (21).

Gram-negative bacteria.

The bacterial cell wall is crucial for bacterial survival. The main component of the bacterial cell wall is peptidoglycan, a heteropolymer composed of cross-linked N-acetylglucosamine and UDP-N-acetylmuramic acid. The synthesis of the bacterial cell wall involves a series of enzymes, such as MurA-G, transglycosylases, and transpeptidases (122, 141). Interrupting any stage in the procedure can lead to failure in peptidoglycan synthesis and subsequent bacteria lysis, and hence, novel antimicrobial agents targeting cell wall biosynthesis are of increasing interest (127). Inhibitors of MurA were discovered by panning against the NEB Ph.D.-12 and Ph.D.-C7C libraries using the Pseudomonas aeruginosa MurA enzyme, with 50% inhibitory concentrations (IC₅₀s) in the 200 μM range reported (87). The same libraries were used to screen the P. aeruginosa MurC enzyme (36). One 12-mer peptide (DHRNPNYSWLKS) and one cyclic 7-mer peptide (CQDTPYRNC) showed similar IC₅₀s (1.5 mM and 0.9 mM, respectively) for inhibition of MurC. However, the peptides are not suitable for therapeutic development because of their low permeability and susceptibility to degradation in plasma. Combined with the mM range of IC₅₀s, the peptide sequences may be valuable only for chemical modification of new antibacterials. Peptide inhibitors for MurE and MurF of P. aeruginosa were identified by panning on the NEB Ph.D.-12 library (96, 97). The two peptides (NHNMHRTTQWPL and TMGFTAPRFPHY) inhibited MurE and MurF activity with IC₅₀s of 500 μM and 250 μM, respectively. Furthermore, inhibitors against P. aeruginosa FtsA and FtsZ, bacterial cell division-related proteins, were screened out from the Ph.D.-12 and Ph.D.-C7C libraries (98, 99). In addition to panning on enzymes, P. aeruginosa whole cells have been chosen as a target. Biosensors immobilized with phage-displayed peptide probes binding to P. aeruginosa whole cells were evaluated and proved to be efficient in diagnostic assays (20).

Helicobacter pylori is a Gram-negative bacterium that causes stomach infections. It can lead to chronic inflammation, which is strongly linked to the development of ulcers and tumors in the stomach. Urease, one of the important virulence factors produced by H. pylori, has been used as a target for phage display (54). By panning on 24-mer and 6-mer phage display peptide libraries, two peptides (TFLPQPRCSALLRYLSEDGVIVPS and YDFYWW) were shown to bind to and inhibit division of H. pylori. The same group selected scFv antibodies that inhibited H. pylori urease from a phage-displayed scFv library (53). Some of the scFvs were displayed on the M13 phage surface, which could be used directly for therapeutic intervention (17, 112); others were selected from a library by biopanning on whole H. pylori cells (118). Whole-cell phage display screening was also used to identify peptides that bound to E. coli. Pini and colleagues reported an antimicrobial peptide (QEKIRVRLSA) with low MIC values (4 to 8 μg/ml) against clinical isolates of multidrug-resistant P. aeruginosa and Enterobacteriaceae (104). In addition, phage probes for Salmonella enterica serovar Typhimurium (131) and Mycoplasma arginini (33), peptides with antimicrobial properties for Haemophilus influenzae (6) and Campylobacter jejuni (5), and an antibody against Chlamydia trachomatis elementary bodies (EBs) (70) have been identified.

Bacterial membranes.

Bacterial membranes are critical for cell survival and function. As the outer leaflet of the bacterial membrane is populated predominately with negatively charged phospholipids, antimicrobial peptides are often comprised of cationic and hydrophobic residues (164). One limitation in the application of phage display to bacterial membrane screening is the paucity of model membrane preparations available. A method of conjugating biotinylated liposomes with phospholipid contents onto magnetic beads coated with streptavidin has been introduced (158). However, the primary disadvantage of this method is that the bound phage cannot be efficiently eluted from the membranes using mild elution conditions, such as detergents and acidic solutions. Tanaka et al. employed bacterial magnetic particles, obtained from Magnetospirillum magneticum strain AMB-1, as a bacterial membrane model for phage biopanning (136). They used phospholipase D (PLD) to disrupt the membrane and successfully recovered the bound phage.

Significant effort has been devoted to targeting lipopolysaccharide (LPS), the major glycolipid molecule present in the outer membrane of Gram-negative bacteria. LPS consists of three subcomponents: (i) a highly variable O-antigen chain, (ii) a core oligosaccharide, and (iii) a relatively conserved phospholipid (lipid A). LPS is an endotoxin that binds the CD14/TLR4/MD2 receptor complex and induces strong responses from the mammalian immune system. Neutralization of LPS is one of the most promising targets for the development of novel antimicrobial agents where phage display could provide an approach for selection of antibodies and peptides specific for LPS. Several antibodies (43, 48, 85) and peptides (44, 59, 60, 78, 92, 169) against LPS from different bacteria strains have been identified. Furthermore, purified lipid A can also be used as a target for phage display screening (137).

Spores.

Spores from microbes such as Bacillus spp. can tolerate extreme environmental conditions and still remain viable. High-affinity scFv antibodies (168) and peptides (62, 140) against native spores of Bacillus spp. were isolated from phage display libraries. These species included Bacillus anthracis, Bacillus cereus, Bacillus subtilis, Bacillus globigii, and Bacillus amyloliquefaciens. In addition, other research was conducted on B. anthracis, a potential bioweapon. The latter study reported several B. anthracis-specific peptides (10, 154) and developed biosensors for detecting the spores of B. anthracis (144).

Viruses.

Several groups worked on epitope mapping and seeking specific antibodies of hepatitis C virus (HCV). Human antibody Fab fragments against HCV NS3 or E2 protein were isolated from phage display libraries by an Italian research group (14, 105). Some of the Fabs against the E2 protein showed nanomolar-range affinity. These Fabs were then converted into full-length IgGs in a CHO cell expression system, which successfully increased the affinity to E2 protein by about 100-fold (12). The same group also performed B-cell epitope mapping of the HCV E2 protein and identified a major epitope with strong neutralization-of-binding (NOB) activity (13). Pereboeva and colleagues established phage display libraries by randomly expressing fragments of HCV NS3 and NS4 and then identified the antigenic regions of those proteins (101). Zhang et al. also used phage display for the epitope mapping of the HCV E2 protein and identified two epitopes involved in the HCV neutralization process (165). In addition to epitope mapping, neutralization peptides for HCV were identified. Human CD81 (hCD81) is considered a receptor for HCV. Its antagonist peptide, SPQYWTGPA, was found by panning on whole hCD81-expressing NIH/3T3 cells (18). Phage display was also used to look for high-affinity antibodies and peptides against hepatitis B virus surface antigen (HBsAg) (51, 152) and hepatitis A virus (HAV) (66) and neutralizing antibodies against hepatitis E virus (HEV) capsid protein (123).

Influenza viruses are known to undergo facile antigenic drift each year, and novel methods are needed to diagnose emergent phenotypes in both seasonal and pandemic influenza. Lim et al. identified H5N1 influenza hemagglutinin (HA)-specific antibodies that can neutralize the H5N1 virus (69). Shortly thereafter, Khurana and colleagues constructed H5N1 gene-fragmented phage display libraries and used them to analyze MAbs and serum samples collected from H5N1-infected individuals. Epitope mapping data derived from this study may contribute to vaccine design and diagnosis of H5N1 infection (58). In addition to influenza H5N1, an antiviral peptide against influenza H9N2 was also identified (108, 109).

There is a large amount of literature on epitope mapping studies using phage display-derived reagents. For example, De Berardinis et al. identified the epitopes of HIV-1 that can prime cytotoxic T-cell responses (34). As well as epitope mapping studies, pathogen detection is a major part of the literature. For example, antibodies against Venezuelan equine encephalitis virus (VEEV) (61) and peptides that recognize the N protein of porcine reproductive and respiratory syndrome virus (PRRSV) (114) have been identified. In 1996, phage display was used for improving the affinity of a MAb that bound specifically to the third hypervariable loop of HIV (138). Furthermore, neutralization antibodies for rabies virus glycoprotein (166) and peptide inhibitors for Andes virus (46) and HIV-1 (151, 161) were identified and showed different mechanisms of inhibition.

Whole viruses have also been employed for phage display screening. Hall et al. used a cyclic phage display library, panned against purified UV-treated Sin Nombre virus (SNV), and obtained peptides with highly specific binding affinity and inhibition of virus infectivity; free peptide showed moderate inhibition, whereas peptides conjugated onto multivalent nanoparticles through a carboxyl linkage resulted in more-potent antiviral activity (45).

Other pathogens.

Phage display has made valuable contributions to research into other important human pathogens, such as malaria (65). In 2001, a 12-mer peptide (PCQRAIFQSICN) that inhibited Plasmodium invasion of midgut and salivary glands was identified (41). In the affinity selection for binding to vector salivary glands, the phage library was injected into the hemocoels of female mosquitoes. After 30 min, the salivary glands were dissected to elute and collect bound phages. To select phages binding to the luminal side of the midgut epithelium, the phage library was fed to mosquitoes. Interestingly, the same peptide was affinity selected in both hemocoel infection and oral ingestion selection strategies. Since then, phage display has rapidly contributed to epitope mapping, diagnosis, therapeutics, and vaccine development in malaria research (22, 29, 143). Da Silva et al. used whole sporozoites of Eimeria acervulina as a target for peptide selection. The peptides they obtained from the NEB Ph.D. library showed high specificity and activity against E. acerulina and Eimeria tenella sporozoites (32). Chen et al. constructed polyclonal Fab phage display libraries against Cryptosporidium parvum (24). These strategies may aid in the development of an effective immunotherapy by targeting antigens that are involved in host-parasite interactions and in the attachment and invasion of pathogens with host cells.

DRUG DISCOVERY APPLICATIONS

The sections above detail the numerous discovery activities and technical advances in the applications of phage display to anti-infective research. Of equal importance and interest is the fate of those antibodies and peptides identified by phage display screening that have progressed to clinical or preclinical development as potential therapeutic drugs. In the following section, we highlight some of the key commercial players in this arena and selected “pathfinder” clinical candidates for infectious diseases (Table 2).

Table 2.

Companies using phage display for discovery of antimicrobials and their lead compounds

Company	Drug name	Pathogen/target	Type of molecule	Molecular target	Developmental stage	Reference(s)
MedImmune (AstraZeneca)	Numax-YTE (MEDI-557)	Respiratory syncytial virus	Humanized IgG1; optimized version of motavizumab	Antigenic A site of the F protein	Phase I	19, 30, 31, 64, 155, 156, 157
Affitech A/S	AT005	Cancer and infectious disease	Expanded-spectrum human version of bavituximab	PS lipid	Preclinical	110, 130
Human Genome Sciences	Raxibacumab	B. anthracis	Recombinant IgG1λ	PA protein	FDA review	82, 86, 89
Phylogica	N/Aa	A. baumannii	Peptide	Whole cells	Discovery	147, 148
Phylogica/MedImmune	N/A	P. aeruginosa	Peptide	Whole cells	Discovery	Unpublished data
Isogenica	N/A	Drug-resistant bacteria	Peptide	N/A	Discovery	93

^aN/A, not applicable.

MedImmune (AstraZeneca).

MedImmune is a pioneer of antibody therapy which expanded through key acquisitions of U.S. Bioscience in 1999 and Aviron in 2002, the integration with Cambridge Antibody Technology (CAT), and subsequent strategic alignment with its new parent company, AstraZeneca, in 2007.

Respiratory syncytial virus (RSV) is a pneumovirus that is a common cause of lower respiratory tract infection in at-risk infants. The humanized MAb palivizumab (Synagis; MedImmune) was developed for RSV prophylaxis of at-high-risk children and has been used since 1998 in the United States and 61 other countries for the prevention of RSV infection. The MAb binds to an epitope in the A antigenic site of the RSV F protein (157). An affinity-optimized version of palivizumab, motavizumab, was generated using a directed evolution approach that allowed manipulation of the binding kinetics of the MAb variants derived from palivizumab (155). In preclinical studies, motavizumab had enhanced activity against RSV. It bound to the RSV F protein 70-fold more than palivizumab and exhibited a 20-fold improvement in neutralization of RSV infectivity (155). In a cotton rat model, motavizumab reduced nasal and pulmonary RSV titers by up to 25- and 100-fold, respectively, and inhibited viral replication in the upper respiratory tract (155). In a phase I study, motavizumab significantly reduced cultivatable RSV in nasal aspirates of previously healthy children with no adverse side effects (64). Subsequently, in a phase III noninferiority trial, motavizumab was compared directly with palivizumab for its safety and rate of RSV hospitalization in preterm infants and children aged less than 24 months with chronic lung disease. Given 15 mg/kg of body weight of either palivizumab or motavizumab monthly, motavizumab was shown to be noninferior to palivizumab, resulting in a 26% relative reduction in RSV hospitalization (19). Motavizumab also demonstrated a significant reduction in outpatient, medically attended lower respiratory tract infections compared with palivizumab (19).

The circulation half-life of palivizumab and motavizumab is 3 weeks, which is not able to cover the RSV season unless several monthly injections are administered (156). Previous studies have shown that the neonatal Fc receptor (FcRn) is important for the long serum half-life of IgG (40, 56). In order to improve the half-life of the MAbs, MedImmune engineered the human Fc region to enhance the affinity to human FcRn through phage display (31). Variants with up to 57-fold increases in affinity were identified in the study, and one of the variants has been further applied for the modification of motavizumab. The engineered antibody, named Numax-YTE or MEDI-557, has shown an increase in affinity of human FcRn of up to 10-fold (30, 156). Numax-YTE is currently in a phase I trial.

Affitech A/S.

Affitech is a newly formed human antibody research company which combined Norway-based Affitech AS and Denmark-based Pharmexa A/S in May 2009. The core technology applied by Affitech A/S is using phage display to generate human antibodies for novel therapeutics. Affitech has created its own antibody phagemid library with a diversity of 10¹⁰ molecules which it is using to develop a series of human antibody therapeutics.

Human antibody AT005, for the treatment of viral diseases, is currently in preclinical research. Affitech is collaborating with Peregrine Pharmaceuticals, a United States-based biopharmaceutical company focused on MAb therapeutics, in developing AT005. The target of AT005 is phosphatidylserine (PS), which is one of the most abundant anionic phospholipid molecules located inside the plasma membrane of healthy cells. When cells become infected by viruses or in the case of solid tumors, PS becomes surface exposed, enabling the viruses and tumors to evade immune recognition. In 2005, a mouse IgG3 MAb (3G4) that bound to the anionic phospholipids of tumors, retarding their growth in multiple mice models of infection, was developed (110). Based on 3G4, Peregrine Pharmaceuticals developed a chimeric version, bavituximab, containing the V_H and V_K domains joined to human IgG_1K constant domains (130). In August 2010 and December 2011, Peregrine Pharmaceuticals completed phase II clinical trials of bavituximab for advanced breast cancer and hepatitis C virus infection, respectively. Human antibody AT005 is an improved, fully human version of bavituximab. Once accepted for development by Peregrine, Affitech will receive clinical development milestone fees and eventual royalties on sales on these expanded-spectrum anti-PS antibody products. However, the status of this antibody is not disclosed on the Affitech website.

Human Genome Sciences (HGS).

HGS is a United States-based pioneer biopharmaceutical company which was founded in 1992. The company is very active in developing protein and antibody therapeutics and owns the first human MAb drug for lupus, belimumab (Benlysta), approved in 2011 by the U.S. Food and Drug Administration (FDA).

Raxibacumab is a human MAb discovered and developed by HGS as a countermeasure for the treatment of inhalation anthrax. It was isolated from a phage display library (licensed by HGS from CAT) by affinity selection against the B. anthracis recombinant protective antigen (PA) protein (82). The isolated antibody was further processed to contain a human IgG1 with a λ light chain and one N-linked glycosylation site per heavy chain (82). Anthrax is caused by the infection of the spore-forming bacterium B. anthracis. As the bacteria can multiply in the host's lungs and produce toxins prior to the onset of symptoms, inhalation anthrax is one of the most fatal forms of infection and, hence, represents one of the greatest threats in biological warfare. The anthrax toxin, produced by actively dividing cells, is a tripartite toxin containing a PA protein, lethal factor (LF), and edema factor (EF). The PA protein is a key facilitator in the progression of anthrax infection. It enhances the binding of LF and EF to the anthrax toxin receptor (ATR) on the mammalian cell surface and promotes LF and EF entering the cells. These toxins inhibit normal immune function, interfere with signal transduction pathways, and ultimately cause cell death. Raxibacumab specifically binds to PA and prevents its binding to ATR with an IC₅₀ of 503 pM (86).

In 2009, two animal efficacy studies demonstrated the life-saving potential of raxibacumab as a prophylactic and postexposure therapeutic agent when administered as a single dose (89). In prophylactic studies, raxibacumab provided complete protection in rabbits and up to 90% survival at day 28 in monkeys at a dosage of 40 mg/kg (86). In postexposure studies, raxibacumab at 40 mg/kg was also protective against the lethal effects of anthrax, with the survival rate significantly higher in rabbits receiving raxibacumab (44%) than in those receiving a placebo (0%). In a monkey therapeutic model, the survival rate increased to 64% during the lethal stage of the challenge compared with the placebo (0%) (82, 86). In 105 healthy volunteers, a randomized, single-blind, placebo-controlled, dose escalation study was conducted to evaluate the safety, pharmacokinetics, and biological activity of raxibacumab. Raxibacumab was safe, well tolerated, and bioavailable after a single intramuscular (IM) or intravenous (IV) dose (86).

Raxibacumab received fast-track designation from the FDA, which enabled expedited development of the drug as a prophylactic countermeasure to natural and artificial infection with B. anthracis. In April 2009, HGS completed 20,000 doses for the U.S. Strategic National Stockpile, and from 2009 to 2012, HGS is delivering another 45,000 doses.

Phylogica and Dynamic Microbials.

Phylogica is a peptide drug discovery company based in Perth, Australia, and Oxford, United Kingdom. The company formed in 2005. They developed Phylomer peptide libraries, which are the main methodology they use to identify peptide drug candidates (147). Phylogica claimed that Phylomer libraries are the largest and most structurally diverse libraries of natural peptides, containing billions of distinct peptide sequences that represent a rich source of biologically active drug leads for a broad range of disease targets. In 2008, Phylogica acquired 100% outstanding shares in Dynamic Microbials, another Australia-based biotechnology company. In the same year, Dynamic Microbials successfully isolated peptides binding to the whole cells of Acinetobacter baumannii and found peptides that inhibited the growth of A. baumannii in vitro using T7 phage display. The peptides were then patented by World Intellectual Property Organization (WIPO) (148). In August 2010, Phylogica initiated the collaboration with MedImmune for seeking new treatments for the control of P. aeruginosa. In this agreement, Phylogica is responsible for screening Phylomer libraries for new leads against P. aeruginosa and MedImmune performs testing and subsequent preclinical and clinical studies.

Isogenica.

Isogenica is a United Kingdom-based company focused on providing protein and antibody discovery and design to pharmaceutical and biotechnology companies. Isogenica developed an in vitro library selection system called CIS display (93). The system can be considered an improved version of phage display. It is a DNA-based approach in which DNA fragments encoding a diverse peptide library are ligated to the gene of a DNA replication initiator protein, RepA. The word “CIS” is derived from cis activity of the RepA protein, which refers to the exclusive binding to its template DNA. The system allows peptides and proteins to be displayed in vitro associated with their encoding DNA templates. In 2010, Isogenica entered into an agreement with the United Kingdom Ministry of Defense agency, the Defense Science and Technology Laboratory (DSTL), regarding the development of antimicrobial peptides against a broad spectrum of bacteria, including MRSA and Clostridium difficile. Isogenica also announced a series of collaboration agreements with pharmaceutical companies and biotechnology companies, such as Johnson & Johnson, Phylogica, and NovaBiotics, for the development of novel anti-infective technology platforms.

Safety, PK, and PD of therapeutic MAbs and peptides.

Pharmacokinetic (PK) and pharmacodynamic (PD) properties of antibody and peptide drugs are more complicated than those of small molecules (molecular weight < 1,000), especially for antibodies because of their massive size. Comprehensive reviews addressing PK and PD properties of protein-based drugs have been published previously (75, 88, 145). The administration of MAb drugs is generally IV, IM, or subcutaneous (SC). Oral administration has not been successfully developed, as antibodies can hardly diffuse through the gastrointestinal epithelium and are easily degraded in the gastrointestinal tract. The distribution of antibody drugs depends on several factors, including the rates of distribution, binding, and elimination in different tissues. Furthermore, antibody drug elimination occurs mainly via secretion into bile (IgA) or intracellular catabolism (IgG). Filtration through the renal system is not important for the elimination of antibody drugs due to their large size. Anti-infective MAbs introduced in this section are mostly in the discovery or preclinical stage. Raxibacumab is the only MAb that is currently under FDA review. Nevertheless, more than 130 proteins or peptides have been approved as therapeutic drugs by the FDA, including more than 20 MAbs and 50 peptides that are therapeutics for different human diseases (1, 67). The number keeps increasing and will hopefully double within the following several years.

PHAGE DISPLAY AND OTHER TECHNOLOGIES FOR DRUG DEVELOPMENT

Like other affinity selection technologies, phage display also has its problems. For novel antimicrobial agent discovery, including peptides and MAbs, phage display showed a high probability of identifying peptides and MAbs with relatively low binding affinity, and a lot of them lack key attributes, such as antimicrobial activity, or need further modification to improve the activity. For example, reconstruction of scFvs and Fabs in mammalian cells is necessary so that full-length antibodies can be achieved. In addition, cationic peptides with antimicrobial properties always show nonspecific binding and a lot of toxicity to mammalian cells (37, 42, 47), which limits the application of phage displays to screen peptide libraries.

In recent years, alternative display technologies, such as aptamers, ribosome display, and mRNA display, are engaged in discovery of novel anti-infectives; however, pros and cons are obvious. For example, aptamer, a single-stranded oligonucleotide that specifically binds to molecular targets such as proteins, benefits from its smaller size, is easier to modify, and has larger-scale synthesis than antibody therapeutics (57, 117). Nevertheless, short half-life and target- and application-dependent optimization are the main drawbacks of it. Moreover, the technologies for generating aptamers, including SELEX (systematic evolution of ligands by exponential enrichment), are largely protected by intellectual property (IP) portfolios (57, 117). Both ribosome display and mRNA display are considered representatives of the next-generation display system. The main advantage of these two technologies is that the cell-free system bypasses the cell culture procedure and makes the screening progress faster and more efficient than cell-based systems (49, 106). In these two systems, protein libraries are displayed in vitro in association with their encoding mRNA. After the affinity selection, mRNA encoding the bound protein is easily achieved and amplified by PCR for next-round screening. The cell-free property also offers benefits in generating and screening large libraries (10¹² to 10¹⁴). Despite the advantages, the two technologies both employed mRNA for encoding libraries, which may degrade quickly. Cell-free systems face downstream processes similar to those of phage display: production of the selected proteins. Although cell-free protein synthesis is an option, it is not sufficient for large-scale production at present.

Compared to in vitro technologies, traditional hybridoma is much more reliable to generate functional MAbs regardless of its expensive cost and long-term process. However, the inherent immunogenicity of rodent antibodies restricts the therapeutic potency of those MAbs. Thus, transgenic mice that express antibodies with human CDRs have been developed (73). The technology has been employed by several biotechnology companies, such as Amgen/Abgenix, Medarex, and Johnson & Johnson, to develop human MAb drugs and has exhibited potential to create very-high-affinity MAbs (72). Recently, a novel technology identifying MAbs from immortalized human B cells was established (139). The European groups successfully isolated anti-SARS coronavirus MAbs by immortalizing memory B cells from a postconvalescent patient. The method definitely provides researchers with an exciting approach for generating therapeutic human MAbs (58, 142).

POTENCY AND DEVELOPMENT TIMESCALES AND SUCCESS

For almost all indications, a drug must bind to its molecular target with a reasonable degree of affinity to be effective. Approximately 60% of small-molecule drug discovery projects fail in hit-to-lead because the biological target was found to not be “druggable” or the molecules did not possess the required potency for the target (25). Phage display may provide advantages in facilitating screening of diverse molecular targets to obtain high-potency (e.g., nM affinity) candidates suitable for therapeutic development within a much faster timescale than that achievable for small molecules (52). In the small-molecule or medicinal chemistry space, initial “hit” molecules from library screening are normally found with target affinities or 50% effective concentrations (EC₅₀s)/IC₅₀s in the 1 to 5 μM range. Many cycles of structure activity design, synthesis, and testing are required to derive molecules with nM potency, a task that often takes more than a year and a large team of dedicated synthetic and medicinal chemists. Indeed, for many targets, requisite potency is simply not achievable with small molecules, which, by definition, possess a limited surface area of contact with the target. In contrast, the power of recombinant techniques can drive selection/affinity maturation cycles using phage display to derive potent “lead” molecules within weeks to months. As a word of caution, it should be noted that high affinities/potencies are not routinely retrieved from commercially available naive libraries (119). Target knowledge and expertise in molecular biology techniques are still required to design a specific peptide library or an immune antibody library that leads to potent, selective hits. In addition, some peptides/MAbs may function only when the peptide is an integral part of the phage coat protein or other avidity scaffold and not when isolated free in solution (104, 129).

It is still true that the vast majority of antibiotics and antivirals on the market or in development are small molecules; however, there is now a nascent pipeline of biological drug candidates in development. Until recently, phage display technologies were used to produce MAbs for application mainly in the areas of cancer and inflammatory disease. Patent disputes impeded broad use of key methods and contributed to the dearth of MAb candidates in the clinic during the 1990s. Screening of naïve and immunized libraries using phage display potentially allows one to develop fully human MAbs against infectious agents and has potential to impact our ability to treat infections refractive to current drug regimes. It is also important to consider the attrition rate of biological versus small molecules in the clinical development phase. The rates of successful transition from phase I (proof of safety) to phase II (proof of efficacy) for all therapeutic MAbs derived from transgenic mouse and from phage display technologies are 87% and 94%, respectively. From phase II to FDA approval, the success rates are 52% and 37%, respectively (91). Hence, phage display-derived MAbs are benchmarking at a 35% rate of successful passage from phase I to launch, compared to an industry-wide average of 12% for a small-molecule drug candidate. Given the paucity of new antimicrobial and antiviral agents in clinical or preclinical development, we believe it is imperative to consider alternative approaches to classical small-molecule discovery.