Recombinant antibodies for diagnostics and therapy against pathogens and toxins generated by phage display

Abstract

Antibodies are valuable molecules for the diagnostic and treatment of diseases caused by pathogens and toxins. Traditionally, these antibodies are generated by hybridoma technology. An alternative to hybridoma technology is the use of antibody phage display to generate recombinant antibodies. This in vitro technology circumvents the limitations of the immune system and allows—in theory—the generation of antibodies against all conceivable molecules. Phage display technology enables obtaining human antibodies from naïve antibody gene libraries when either patients are not available or immunization is not ethically feasible. On the other hand, if patients or immunized/infected animals are available, it is common to construct immune phage display libraries to select in vivo affinity‐matured antibodies. Because the phage packaged DNA sequence encoding the antibodies is directly available, the antibodies can be smoothly engineered according to the requirements of the final application. In this review, an overview of phage display derived recombinant antibodies against bacterial, viral, and eukaryotic pathogens as well as toxins for diagnostics and therapy is given.

Abbreviations

CDC

Centers for Disease Control and Prevention

DENV

dengue virus

DIVA

differentiating infected from vaccinated animal

ET

edema toxin

LF

lethal factor

MSP‐3

merozite surface protein 3

PA

protective antigen

SEB

staphylococcal enterotoxin B

TTX

tetrodoxin

VEEV

Venezuelan equine encephalitis virus

WEEV

Western equine encephalitis virus

1. Introduction

Antibodies are essential molecules as tools for basic research 1, diagnostics 2, and therapy 3. In the past—and still today—polyclonal antibodies were produced as serum in animals like horses 4. A milestone in antibody generation was the development of hybridoma technology that allows the production of monoclonal antibodies 5. But the hybridoma technology has drawbacks like limited number of candidates, possible instability of the aneuploid cell lines 6, inability to provide antibodies against highly conserved antigens and most of all its limited application to generate human antibodies 7. The hybridoma technology essentially gives murine antibodies that can be used for diagnostic or research purposes. However, their therapeutic applications are limited because repeated administration of murine antibodies can cause human anti‐mouse antibody reaction, reducing antibody half‐life and leading to severe side effects such as anaphylactic shock 8. A strategy to circumvent these problems is antibody humanization or the use of transgenic animals in which the original antibody gene repertoire is replaced with a human gene repertoire 9, 10, 11, 12. Another strategy is the human hybridoma technology resulting in human antibodies 13, 14, although this technology has the same problems of the murine hybridoma technology regarding the limitation of the immune system.

A technology that overcomes the limitation of the immune system is antibody phage display. Since this technology involves an in vitro selection process, it is completely independent of any immune system. The display method most commonly used today is based on the work of Georg P. Smith on filamentous phage, which infect E. coli 15. The selection process is called “panning,” referring to the gold digger's tool 16.

M13 phage display technology was further developed 1990/91 for antibodies in three places in parallel: Heidelberg (Germany), Cambridge (UK), and La Jolla (USA) 17, 18, 19, 20. Two different genetic systems have been developed for the expression of the antibody::pIII (phage protein III) fusion proteins for phage display. First, the antibody genes can be directly inserted into the phage genome fused upstream of the wild‐type pIII gene 20. Second, most successful systems, uncouple antibody expression from phage propagation by providing the genes encoding the antibody::pIII fusion proteins on a separate plasmid (called “phagemid”) that contains a phage morphogenetic signal for packaging the vector into the phage particles 18. No matter the used phage display system, antibody fragments are displayed on the surface of M13 phage and the corresponding antibody gene is packaged in the phage particle. Other phage display technologies using phage lambda 21 or T7 22 are less suitable for the display of antibody fragments. In addition, these phages are lytic phage that aggravates the practical work. The most common antibody formats used for antibody phage display are the single chain fragment variable (scFv) 23, 24, 25 or fragment antigen binding (Fabs) 26, 27. Other antibody formats used for phage display are single chain Fabs (scFab), human VH domains (dAbs), the variable domains of camel heavy chains (VHHs) and immunoglobulins of sharks (IgNARs) 28, 29, 30, 31, 32, 33. Figure 1 shows an antibody phage and different antibody fragments. For veterinary research, chicken libraries are often used 34, 35. In contrast to what occurs in humans, the diversity of chicken antibody genes is the result of gene conversion, and the N‐ and C‐terminal parts of chicken's VH and VL are always identical, facilitating antibody gene amplification, and library cloning 36, 37. Also rabbit phage display was used to generate antibodies against pathogens 38.

Figure 1.

(A) Schema of an antibody (scFv) phage. (B) Different antibody formats used for phage display. Abbreviations: Fab, Fragment antigen binding; Fc, Fragment constant; Fv, Fragment variable; pIII, pVI, pVII, pVIII, pVIX, phage protein III, VI, VII, VIII, VIX; scFab, single chain Fab; scFv, single chain fragment variable; VH, variable domain heavy chain; VHH, variable domain of the camel heavy chain.

In the selection process (panning), the antigen is immobilized to a solid surface, like column matrixes 18, magnetic beads 39, or plastic surfaces with high protein‐binding capacity such as polystyrene tubes or respectively microtitre wells, which is the most common method 40. An other option is the panning in solution using biotinylated antigens followed by a “pull‐down” with streptavidin beads 41. The vast excess of nonbinding antibody phage will be removed by stringent washing. Subsequently, the bound antibody phage will be eluted, e.g. by pH shift or trypsin, and reamplified by infection of E. coli. After infection of the phagemid bearing E. coli with a helperphage, new antibody phage will be produced. The selection cycle will be repeated and the number of antigen‐specific antibody phage clones should increase with every panning round. Usually 2–3 panning rounds are performed. Finally, monoclonal antibody phage or monoclonal soluble antibodies can be identified by, e.g. ELISA 42, immunoblot 40, or flow cytometry 43. The antibody fragment genes can be subcloned into any other antibody format, e.g. scFv‐Fc or IgG 23, 27, 42, 44, 45. A schema of the selection process is given in Fig. 2.

Figure 2.

Schema of antibody selection using phage display.

Regarding the source of genes, antibodies can be selected from two types of libraries: immune libraries and universal libraries. Immune libraries are constructed from immunized/infected donors and typically used in medical research to obtain an antibody against a particular target antigen, e.g. an infectious pathogen like Ebola virus 46. An advantage of this kind of library is that the V‐genes contain hypermutations and are affinity matured, although its development can be restricted to ethical constraints.

The alternative are universal or “single pot” libraries, which includes naïve, semisynthetic and synthetic libraries that are designed to isolate antibody fragments binding to every possible antigen, at least in theory 44, 47. Naïve libraries are constructed from rearranged V genes from B cells (IgM) of nonimmunized donors. Examples for this library type are the naïve human Fab library constructed by de Haard and colleagues 26 and the HAL scFv libraries 23, 48. Semisynthetic libraries are constructed from unrearranged V genes from pre‐B cells (germline cells) 49 or from one antibody framework 50 in which one or several CDRs, but always the CDR H3, are randomized. Often used semisynthetic libraries are the Tomlinson I and J libraries using one defined framework VH3‐23 and Kappa IKV1‐39 with randomized CDR2 and CDR3 51.

A combination of naïve and synthetic repertoire was used for the FAB310 antibody gene library. In this library, light chains from autoimmune patients were combined with a Fd fragment (VH+CH1) containing synthetic CDR1 and CDR2 in the human VH3‐23 framework and naïve CDR3 regions, originated from autoimmune patients 27. Fully synthetic libraries are made of human frameworks with randomized CDR cassettes 52, 53, 54. The theoretical size of these universal libraries is usually higher than 10¹⁰ independent clones 24, 48, 54, 55, 56.

To date, 53 antibodies and antibody conjugates were approved by EMA and/or FDA (status January 2016) (http://www.imgt.org/mAb-DB/query.action, Development status: Phase M in search field) and about 350 antibodies were under development in 2013 57. Most approved therapeutic antibodies are for cancer and autoimmune diseases and the annual sales of therapeutic antibodies exceeded 50 billion US$ in 2013 58. The mechanisms of therapeutic antibodies are manifold and include neutralization of substances, e.g. toxins 59 or cytokines like tumor necrosis factor alpha 60, blocking of receptors like epidermal growth factor receptor 61, binding to cells and modulating the host immune system 62, or combinations of these effects 63. Currently, two recombinant antibodies are approved for the treatment of pathogens or toxins. Raxibacumab is a human antibody for anthrax treatment derived from a phage display library from Cambridge Antibody Technology (now Medimmune, part of AstraZeneca) in cooperation with Human Genome Science (now GlaxoSmithKline) 64. The antibody palivizumab for the treatment of Respiratory syncytial virus bronchiolitis is a classical humanized antibody 65. A further antibody, but also not derived from phage display, the Clostridium difficile antibody bezlotoxumab is in clinical phase 3 66.

An overview of recombinant antibodies derived from phage display against bacterial and viral pathogens, eukaryotic pathogens (parasites, fungi), and toxins as well as detailed examples for diagnostic and therapy are given in the next sections.

2. Recombinant antibodies against bacteria

The most therapeutic antibodies against bacterial targets are generated against toxins. These antibodies are described in the section “recombinant antibodies against toxins.” The majority of antibodies against bacteria are developed in order to facilitate diagnostics in patients 67, 68 and environmental samples 69, 70. In general, cultural and microbial detection of bacteria is regarded as standard in diagnostics for many pathogens, e.g. Mycobacterium tuberculosis 71 or Salmonella Typhimurium 72. Since these methods are often time‐consuming and require experienced lab personal, high throughput analysis is often limited. Real‐time PCR measurements have been developed for the diagnostics of many bacteria 73, offering high sensitivity and specificity of detection. But sample treatment is still needed in many cases, including expensive and complex laboratory devices 74. An other approach is MS for diagnostic, but this technology needs expensive devices 75. Antibody‐based diagnostic like ELISA would be more simple and easy to use, also in developing countries.

In order to generate antibodies with the desired‐binding properties, phage‐display has been used to select antibodies on proteins or polysaccharides of Chlamydophila psittaci 76, Chlamydia trachomatis 77, Haemophilus influenzae 78, Listeria monocytogenes 79, Mycobaterium bovis 35, Mycobacterium tuberculosis 68, 80, 81, Porphyromonas gingivalis 67, Ralstonia solanacearum 69, Salmonella Typhimurium 2, 82, and Yersinia pestis 83. But even selections on cell lysate or whole cells were performed on Mycobacterium avium 84, Bacillus anthracis 70, 85, Moraxella catarrhalis 86, Lawsonia intracellularis 87, Lactobacillus acidophilus 88, Helicobacter pylori 89, Brucella melitensis 90, and Bordetella pertussis 91. In the following paragraphs, we give detailed examples for antibody generation using phage display and antibody engineering against different bacterial pathogens.

Tuberculosis (TB) is a bacterial infection caused by Mycobacterium tuberculosis (Mtb), a pathogen with particular cell wall and membrane characteristics that is able to infect and multiply within alveolar macrophages leading to cough, weakness, and fever 92. In 2013, 9 million people were infected and 1.5 million died worldwide 93. In order to improve treatment and control disease spread, TB must be diagnosed as early as possible. One prominent target for this purpose is the bacterial Antigen 85 (Ag85) that is the most abundant protein secreted by Mtb 94. It is a complex of three highly homolog proteins (Ag85A, Ag85B, and Ag85C) with a molecular size of 30–32 kDa each 95. In a novel approach by Ferrara et al. 68 monoclonal antibodies were selected against the Ag85 complex by combining phage‐ and yeast‐display. First, a naïve scFv‐library was preselected against Ag85 by phage display, reducing the diversity from ∼10¹¹ unique scFv clones to ∼10⁵ colony forming units. Second, the enriched sublibrary was cloned into a yeast‐display system. Each clone of the selection output was screened individually for antigen binding by FACS. Those with the highest antigen‐binding signals were sorted and further analyzed. In total 192 clones were sequenced revealing 111 genetically unique scFv clones. Further screening assays identified seven antibody pairs, which can detect Ag85 down to a concentration of 6.1 nM in the absence of serum and 22.7 nM in its presence. Interestingly, none of the antibodies were absolutely specific for one of the Ag85 subunits of the complex. All three subunits A, B, and C were bound although not equally. In contrast, Fuchs et al. selected five human antibodies against the recombinant Ag85B protein from the naïve HAL7/8 libraries 81. Three of them bound specifically to the 85B protein in ELISA (LOD: 5 ng/mL). Even sandwich detection of recombinant 85B in ELISA (LOD: 10 ng/mL) and integration into lateral flow immuno assay (LOD: <5 ng/mL) was shown. But antigen detection in Mtb cell extracts or culture filtrates was restricted to direct ELISA and immunoblot assay. The authors of both papers argued that the selected antibodies must be affinity matured in order to reach required detection limits. Successful implementation of this concept was demonstrated by Sixholo et al. 80. Chicken antibodies were selected from the semisynthetic Nkuku library against a 16 kDa recombinant antigen of M. tuberculosis, also known as heat shock protein X (HspX). After four rounds of panning, three scFvs were identified in ELISA. The clone with the lowest ELISA signal was selected for in vitro engineering using two different approaches. First, a mutant library was generated from the parental scFv by error‐prone PCR with a diversity of ∼3 × 10⁷ unique scFv. This library was selected again on the HspX antigen under more stringent conditions, leading to identification of three mutant scFvs. These contained one or three individual amino acid exchanges, occuring both in CDR and framework regions. All mutants showed increased ELISA‐binding signals, reaching a signal ∼11 times higher when compared to the parental scFv. Furthermore, the mutants showed improved association and dissociation kinetics in ELISA and SPR analysis. In the other approach, the length of the scFv linker was reduced from 15 amino acids to a single glycine residue, forcing tetramerization of antibodies. Due to cooperative binding, the apparent functional affinity was improved. In ELISA, tetrameric scFv generated a ∼14 times higher ELISA signal, and SPR proved increased association and decreased dissociation rates.

Porphyromonas gingivalis is one of the major pathogens involved in periodontitis 96. Periodontitis is an inflammatory disease, which causes the loosening and loss of teeth. Secreted cysteine proteases like RgpB contribute to disease pathology and represent potential biomarkers for disease detection and progression 97. Skottrup et al. 67 generated a naïve antibody library from camels with a diversity of ∼5 × 10⁷ clones. A special feature is that the antibodies consist of a single monomeric VHH domain, also called nanobodies. Advantages of this format are the small size (∼15 kDa), the ease production in E. coli, and the convex paratope that is formed which enables the targeting of cryptic epitopes. The library was selected on immobilized RgpB. One clone was isolated that binds specifically to cell surface displayed and soluble RgpB with an affinity of 362 pM. A detection limit of ∼8 × 10⁶ cells/mL of saliva was reached when tested by subtractive inhibition ELISA. But catalytic activity of RgpB was not inhibited by antibody binding.

Salmonella Typhimurium, is one of the most important pathogens of foodborne gastrointestinal infections 98. Meyer et al. 82 first identified novel immunogenic proteins of S. Typhimurium from a genomic library using oligopeptide phage display. In a second step human antibodies were selected against these targets from the naïve HAL gene libraries. For DIVA (differentiating infected from vaccinated animals) vaccine development, S. Typhimurium strains lacking a marker protein, e.g. OmpD (outer membrane protein D) were developed (DIVA vaccines) 99. To allow the discrimination between vaccinated animals and infected animals a diagnostic of this specific marker is necessary. Here, Meyer et al. 2 generated scFvs against OmpD for diagnostics purposes. Only some of the generated antibodies were suitable for a competitive ELISA using swine serum showing also the difficulties when developing diagnostic assays that are often hampered by the complexity of serum samples.

An overview of recombinant antibodies generated by phage display against bacterial pathogens is given in Table 1.

Table 1.

Recombinant antibodies derived by phage display against pathogenic bacteria

Bacteria	Target	Library type	Antibody format	Antibody origin	Application	Reference
Bacillus anthracis	S‐layer protein EA1	Immune	VHH	Llama	ELISA, immunoblot	85
Bacillus anthracis	Live bacteria	Immune	scFv	Mice	ELISA, immunoblot, IF	70
Bordetella pertussis	Filamentous hemagglutinin, pertactin	Immune	scFv	Mice	ELISA, in vitro inhibition, in vivo studies	91
Brucella melitensis	Radiated bacteria	Immune	scFv	Mice	ELISA	90
Chlamydophila psittaci	2.4[2.8]2.4‐linked Kdo tetrasaccharide	Immune	scFv	Mice	ELISA, IF	76
Chlamydia trachomatis	Elementary bodies	Naïve	scFv	Human	ELISA, immunoblot, IF	77, 217
Clostridium difficile	Different surface proteins (including FliC and FliD)	Semisynthetic	scFv	Human	ELISA, immunoblot, in vitro motility assay	218
Haemophilus influenzae	Capsular polysaccharide	Immune	Fab	Human	ELISA	78
Helicobacter pylori	Bacterial lysate	Immune	scFv	Human	ELISA	89
Lactobacillus acidophilus	S‐layer protein	Naïve	scFv	Human	FACS, immunoblot	88
Lawsonia intracellularis	Live bacteria	Semisynthetic	scFv	Human	ELISA, IF	87
Listeria monocytogenes	Internalin B	Immune/naïve	VHH	Alpaca, llama, camel	ELISA	79
Moraxella catarrhalis	HMW‐OMP	Semisynthetic	scFv	Human	ELISA, FACS, immunoblot, in vitro inhibition	86
Mycobacterium avium	Cell lysate	Immune	scFv	Sheep	ELISA, immunoblot, FACS, IF	219
Mycobaterium bovis	HSP65	Semisynthetic	scFv, scFv‐IgY‐Fc	Chicken	ELISA, immunoblot	35
Mycobacterium tuberculosis	Antigen 16 kDa (HspX)	Semisynthetic	scFv	Chicken	ELISA	80
Mycobacterium tuberculosis	Antigen 85B	Naïve	scFv, scFv‐Fc	Human	ELISA, immunoblot, lateral flow strip assay	81
Mycobacterium tuberculosis	Antigen 85	Naïve	scFv, scFv‐Fc	Human	ELISA	68
Porphyromonas gingivalis	RgpB	Naïve	VHH	Camel	ELISA	67
Ralstonia solanacearum	LPS	Naïve	scFv	Human	ELISA, IF, immunoblot	69
Salmonella Typhimurium	OmpD	Naïve	scFv	Human	ELISA	2
Salmonella Typhimurium	5 different immunogenic proteins	Naïve	scFv	Human	ELISA	82
Yersinia pestis	F1	Naïve	scFv	Human	ELISA	83

ELISA, enzyme‐linked immunosorbent assays; IHC, immuno histo chemistry; FACS, fluorescence‐activated cell sorting.

This article is being made freely available through PubMed Central as part of the COVID-19 public health emergency response. It can be used for unrestricted research re-use and analysis in any form or by any means with acknowledgement of the original source, for the duration of the public health emergency.

3. Recombinant antibodies against viruses

Up to now, a large panel of antibodies against various viruses has been generated from either naïve or immune libraries using phage display technology. Panning against peptides, recombinant viral proteins, or complete virus particles has led to the identification of antibodies directed against human pathogenic viruses such as Sin nombre virus 100, Dengue virus 101, 102, Influenza virus 103, 104, VEEV 105, Norovirus 106, SARS coronavirus 107, or Hepatitis C 108 from naïve antibody gene libraries. Other antibodies were selected from immune antibody gene libraries targeting, e.g. Western equine encephalitis virus (WEEV) 109, HIV 110, 111, SARS 112, Yellow fever virus 113, or Influenza virus 114, 115. Semisynthetic libraries were also used to generate antibodies specific for Influenza virus 116. Beside human and animal viruses, antibodies were also generated against plant viruses 40, 117, 118.

Libraries originated from different species have been successfully employed to isolate virus‐specific antibodies in the past such as those from macaque 119, mouse 120, chimpanzee 121, llama 122, chicken 123, and human origin 124.

Most of the virus‐specific antibodies have been isolated from libraries in scFv format 125, 126, although Fab 127, 128 and VHH libraries 122 were also successfully used. An interesting approach was used by Xiao et al. using the antibody CH2 domain as scaffold to generate binders against gp120 of HIV 129.

Different antibody characteristics have an influence on virus binding and neutralization. Neutralizing antibodies prevent cell binding of the virus. The anti‐gp41 antibody HK20 has a higher neutralizaton rate as scFv or Fab compared to IgG showing, that the epitope is less accessable for the IgG format 130. Another interesting example is the anti‐gp41 VHH 2H10. Here, a tryptophan in the CDR3 is not relevant for epitope binding, but essential for virus neutralization 131.

In the following paragraphs, we give detailed examples for antibody generation using phage display and antibody engineering against different virus groups.

Vaccinia virus is the prototype virus in the genus of Orthopoxvirus. It is a relatively large DNA virus with a genome of about 200 kbp 132. The genus Orthopoxvirus includes various species such as monkeypox virus, cowpox virus, and especially variola virus that is the causative agent of smallpox in humans. Naturally occurring smallpox has been eradicated in 1977 because of a massive WHO vaccination program that began in 1967. However, no vaccination of the civilian population is conducted nowadays. The potential threat of intentional release has renewed the search for safe and effective smallpox vaccines as case fatality rates of 30% or more among unvaccinated subjects are reported 133. Because orthopoxviruses are highly related, it is assumed, that immunity against one poxvirus goes along with immunity against most members of the entire virus family 132, 134. Using an immune scFv phage display library constructed from vaccinia virus immunized patients, a panel of human vaccinia‐specific antibodies was selected. Plaque‐reduction neutralization tests revealed that seven of these antibodies neutralized vaccinia as well as cowpox virus in vitro. Five of those antibodies additionally neutralized monkeypox virus 124. Other antibodies were generated from a Fab immune library derived from vaccinia virus immunized chimpanzee. Converted into a chimeric chimpanzee/human IgG format, two antibodies displayed high affinities to vaccinia protein B5 (K_d of 0.2 and 0.7 nM). Antibody 8AH8AL was neutralizing in vitro for vaccinia and smallpox virus and proofed to be protective in mice challenged with vaccinia virus even when administered 2 days after challenge. In this model 8AH8AL proved to provide significantly greater protection than that of the previously isolated rat anti‐B5 antibody 19C2 135. Vaccinia had to be used as model, because the final confirmation of protection against smallpox is not possible.

Ebola virus and Marburg virus, two filoviruses, cause severe hemorrhagic fever and possess high mortality of up to 90% in humans. In addition to public health concerns associated with natural outbreaks, Ebola virus might be a potential agent of biological warfare and bio‐terrorism 136. Human antibodies directed against Ebola virus were selected from a library originated from patients that recovered from infection in the 1995 Ebola virus outbreak in Kikwit, Democratic Republic of Congo 137. Several antibodies against various viral proteins such as nucleoprotein, envelope glycoprotein, and secreted envelope glycoprotein have been isolated in this study. One antibody (KZ52), specific for envelope glycoprotein, was able to neutralize in vitro as Fab (50% neutralization at 0.4 μg/mL) and as full IgG (90% neutralization at 2.6 μg/mL) 46. Follow‐up studies were showing effective protection in vivo in a Guinea pig Ebola challenge model when the antibody was administered up to one hour postviral challenge 138. Interestingly, KZ52 was not protective in macaques challenged with Ebola even if the antibody was given as a two‐dose treatment with the first dose one day prior viral challenge and the second dose 4 days postchallenge 139. A murine scFv and two shark IgNAR V immune libraries were generated against inactivated Zaire Ebola virus to yield various antibodies specific for the viral matrix protein VP40 and the viral nucleoprotein 136. Interestingly, this work represents the first example of a successful targeted IgNAR V isolation from a shark immune response library.

Dengue virus (DENV), a member of the Flaviviridae family, is responsible for at least 100 million symptomatic infections each year and became a major health and economic burden in over 50 countries worldwide 140, 141, 142. It is a positive strand RNA virus with a ∼11 kb genome, that compromise a single open reading frame. The four circulating serotypes of dengue virus show approximately 70% sequence homology 128, 141. Fab monoclonal antibodies to dengue type 4 virus were isolated from a chimpanzee immune library. Two Fab, namely 5H2 and 5D9 neutralized DENV‐4 efficiently with a titer of 0.24–0.58 μg/mL by plague reduction neutralization test 143. Another study selected human scFv antibodies specific to dengue virus envelope protein by panning against recombinant full‐length envelope protein and its domain III 102. Because DENV envelope protein is an essential molecule for virion assembly and virus entry, scFvs selected in this study were shown to exhibit inhibitory effects on DENV infection in vitro 102. Dengue nonstructural protein 5 (NS5) is essential for viral replication and host immune response modulation, making it an excellent target for dengue‐inhibiting antibodies. A naïve human Fab‐phage library was screened for NS5‐specific antibody fragments using various NS5 variants from Dengue Virus serotypes 1–4 as antigens for panning and characterization 128. Using NS5 from alternating dengue serotypes for each round of panning, this strategy resulted in the identification of two clones that are cross‐reactive against all four dengue serotypes. Another study selected antibodies using phage display by panning with Dengue virus particles directly captured from supernatant of infected Vero cells. Here, highly serotype‐specific antibodies were generated. From a total of nine antibodies, seven were shown to be specific to only one serotype. One Dengue‐3 selected clone cross‐reacted with Dengue 1, whereas another clone showed cross‐reactivity with all serotypes despite being selected solely on Dengue 2 particles. Interestingly, all of the obtained antibodies recognized several strains of distinct genotypes within the corresponding serotype 101. Panning against dengue envelope protein resulted in the identification of an antibody (C9) from a mouse/human chimeric Fab library that cross‐reacts with DENV1‐3 and neutralizes DENV2 in cell‐based assays after conversion into full‐length IgG 141. Besides scFv and Fab, variable domain heavy‐chain antibodies (VHH antibodies) were also selected using phage display technology. After four rounds of panning on recombinant DENV 2 NS1 protein, 20 positive clones were selected. Affinity measurements with NS1 revealed a K _D value of 2.79 × 10⁻⁸ M for the best VHH antibody P2 122.

Venezuelan equine encephalitis virus (VEEV), an alphavirus of the Togoviridae family, causes equine epidemics but can also cause encephalitis in humans 144, 145. Because this virus is classified as Category B agent by the Centers for Disease Control and Prevention (CDC), much research has been done to generate neutralizing antibodies against it. Antibodies were generated from an immune library from human donors targeting both VEEV envelope glycoproteins E1 and E2 146. The isolated Fabs L1A7 and F5 were neutralizing in vitro, with F5 being 300 times more effective than L1A7. Subsequently, F5 was converted into full IgG format and was employed to generate neutralization‐escape variants of VEEV for epitope mapping. Within another study, an immune macaque library was used to generate human‐like antibodies 119. One of these antibodies, scFv‐Fc ToR67‐3B4, was protective in mice when administered 6 h postviral challenge with VEEV Trinidiad strains, showing 80–100% survival after a challenge with 100‐fold LD₅₀. However, scFv‐Fc ToR67‐3B4 was not able to neutralize Trinidad strain in vitro, but other VEEV strains instead, showing that neutralization is not mandatory for an in vivo protective antibody 119. Another study describes the selection of antibodies from a human naïve scFv gene library using complete, active VEEV particles as antigen. In this case, specific detection of the VEEV strains TC83, H12/93, and 230 by the selected antibodies was proven. Remarkably, none of the selected scFv phage clones showed cross‐reactivity with Alphavirus species of the Eastern equine encephalitis virus and WEEV antigenic complex or with Chikungunya virus, making them ideal tools for the immunological detection and diagnostic of Alphavirus species 105. Two different scFv antibody libraries were constructed from WEEV immunized macaques. Subcloned as scFv‐Fc, three antibodies from these libraries specifically bound WEEV in ELISA with little or no cross‐reactivity with other alphaviruses and were found to be neutralizing in vitro. In this study, the first antibodies against WEEV, that were shown to be neutralizing in vitro, were developed. About 1 ng/mL of the best antibody (ToR69‐3A2) neutralized 50% of 5 × 10⁴ TCID50/mL WEEV 109.

Flu is a disease caused by influenza viruses. In the last years, the bird flu (H5N1) and the pandemic swine flu (a variant of H1N1) moved in the research focus. Due to many genetic events, such as antigenic drift and shift, new influenza variants will occur in the future and will challenge vaccine and diagnostic development 147, 148. Different groups developed antibodies by phage display against influenza viruses. Sui et al. 104 selected antibodies from a nonimmune scFv library against the H5 hemagglutinin ectodomain. Hemagglutinin is a trimer and the extracelluar part consists of a stalk domain and a globular head domain 149. They identified ten antibodies binding to the trimeric H5, but not monomeric H5. Interestingly, nine of these antibodies shared the same germline framework (VH1‐69). These antibodies were converted into IgG1 and were protective in mice with 10 or 15 mg/kg in a prophylactic or therapeutic challenge model, respectively. Very remarkably, some antibodies cross‐neutralized H1, H2, H5, H6, H8, and H9 influenza strains. These are phage display derived antibodies that are candidates for broad‐spectrum influenza immunotherapy 104.

Rabies is caused by the rabies virus that infects the central nervous system. Before Louis Pasteur developed the rabies vaccination, the disease was always fatal. The current postexposure therapy is based on vaccination—as performed by Pasteur in 1885—and polyclonal anti‐rabbies immunoglobulins 150, 151, 152. One hundred forty‐seven unique recombinant antibodies against the rabies glycoprotein were selected from two immune scFv libraries 153. The neutralization of 27 street rabies viruses was tested in vitro and the best neutralizing antibodies were tested in vivo in a hamster rabies infection model. Here, the antibody CR4098 showed 100% postexposure protection with 40 IU/kg 154. This antibody was further analyzed in combination with an other human antibody CR57, derived by somatic cell hybridization technique 155, in in vitro and in vivo models 152. The safety of this mAb cocktail named CL184 was tested in a clinical phase 1 study 156 and subsequently in phase 2 studies. The antibodies were named Foravirumab (CR4098) and Rafivirumab (CR57).

An overview of recombinant antibodies generated by phage display against viruses is given in Table 2.

Table 2.

Recombinant antibodies derived by phage display against viruses

Virus	Target	Library type	Antibody format	Antibody origin	Application	Reference
Blue tongue virus	Complete virus	Semisynthetic	scFv	Chicken	ELISA	34, 123
Cucumber mosaic cucumovirus	Complete virus	Semisynthetic	scFv	Human	ELISA, immunoblot	117
Dengue	Dengue virus envelope protein E	Naïve	scFv	Human	ELISA, Immunofluorescent assay, in vitro neutralization	102
Dengue	Dengue NS5 protein	Naïve	Fab	Human	ELISA, immunoblot, dot blot	128
Dengue	Dengue virus envelope protein	Naïve	Fab, IgG	Human, mouse (panel of hybridoma clones)	ELISA, immunoblot, IHC	141
Dengue	Dengue NS3	Naïve	Fab	Human	ELISA, in vitro neutralization	142
Dengue	NS1 protein	Naïve (nonimmune)	VHH	Llama	ELISA, lateral flow immunochromatograpic assay	122
Dengue	n.d.	Immune	Fab, IgG	Chimpanzee	ELISA, immunoprecipitation, in vitro neutralization	143
Ebola	Nucleoprotein, envelope glycoprotein, secreted envelope glycoprotein	Immune	Fab, IgG	Human	ELISA, immunostaining, immunoprecipitation, in vitro neutralization	46
Ebola	Nucleoprotein	Synthetic	scFv, IgG	Human	ELISA, immunoblot	220
Ebola (Zaire)	Viral matrix protein VP40, nucleoprotein	Immune	scFv, IgNAR	Mice, shark	ELISA, immunoblot	136
Epstein–Barr virus	LMP1	Naïve	Fab	Human	ELISA, immunoblot, IF, FACS, in vitro inhibition	221
Foot‐and‐mouth disease virus (FMDV)	3ABC	Immune	scFv	Chicken	ELISA, immunoblot	222
Grapevine leafroll‐associated virus 3	Coat protein	Immune	scFv	Mice	ELISA	118
Grapevine virus B	Virus particles	Semisynthetic	scFv	Human	ELISA	223
Hantavirus	Nucleoprotein	Immune	VHH	Llama	ELISA, immunoblot	224
HCMV	Gycoprotein B and H	Immune	scFv	Human	ELISA, in vitro neutralization	225
Hendra and Nipah virus	Attachment envelope glycoprotein G	Naïve	Fab, IgG	Human	ELISA, immunoprecipitation, Immunoblot, in vitro neutralization	226
Hepatitis A	Hepatitis A Capsid	Immune	Fab, IgG	Chimpanzee	ELISA, in vitro neutralization	227
Hepatitis A	HBsAG	Naïve	scFv	Human	ELISA	228
Hepatitis C	NS5B	Naïve	scFv	Human	ELISA, IF, in vitro neutralization	229
Hepatitis C	Core, E1, E2	Immune	scFv	Human	ELISA	230
Hepatitis C	Core protein	Immune	Fab	Human	ELISA	231
Hepatitis C	E2 glycoprotein	Immune	Fab	Human	ELISA	232
Hepatitis E	ORF2 protein	Immune	Fab	Chimpanzee	ELISA, immunoblot	233
Herpes simplex virus	HSV1, ‐2 lysate	Immune	Fab	Human	In vitro inhibition, immuno precipitaton	234
Herpes simplex virus	Glycoprotein gD, gB	Presumably immune	Fab	Human	ELISA, immuno precipitation	235
Herpes simplex virus	Virus lysate?	Immune	Fab	Human	ELISA, IF	236
HIV‐1	Integrase	Immune	scFv	Rabbit	ELISA, immunoblot, IF, in vitro neutralization	38
HIV‐1	gp140	Immune	scFv, scFv‐Fc	Human	ELISA, immunoblot, immunoprecipitation, in vitro neutralization	111
HIV‐1	gp140	Immune	Fab, IgG	Human	ELISA, immunoblot, in vitro neutralization	237
HIV‐1	gp140	Immune	VHH	Llama	ELISA, in vitro neutralization	238
HIV‐1	gp120	Synthetic	CH2 domain	Human	ELISA, in vitro neutralization	129
HIV‐1	p24	Immune	scFv	Mouse	ELISA	110
HIV‐1	gp41	Synthetic	Fab	Human	Immunoblot, in vitro neutralization	239
HIV‐1	gp41	Naïve	scFv	Human	ELISA, in vitro neutralization	240
HIV‐2	gp125 protein	Immune	Fab	Human	ELISA, in vitro neutralization	241
Human metapneumovirus	F ectodomain	Presumably immune	Fab	Human	ELISA, IF, in vitro neutralization, in vivo protection	242
Infectious haematopoietic necrosis virus	n.d.	Naïve (nonimmune)	scFv	Mouse	ELISA, immunoblot, IHC	120
Influenza A	HA (stem region)	Semisynthetic	scFv	Human (IGHV1‐69)	ELISA, in vitro neutralization	116
Influenza A	HA ectodomain	Naïve	scFv	Human	ELISA, flow cytometry, immunoprecipitation, in vitro neutralization, in vivo protection	104
Influenza A	HA (stem region)	Presumably immune	Fab	Human	ELISA, in vitro neutralization	127
Influenza A (H5N1)	HA	Semisynthetic	scFv	Human	ELISA	126
Influenza A (H5N1)	HA	Naïve	scFv	Human	ELISA, in vitro neutralization, in vivo protection	243
Influenza A (H5N1)	HA	Immune	Fab	Chicken	IF, in vitro neutralization, immunoblot	244
Influenza A (H5N1)	HA cleavage site	Immune	Fab	Mice	ELISA, IF	245
Influenza A (H5N1)	NS1	Naïve	scFv	Human	ELISA, in vitro neutralization, IF	246
Japanese encephalitis virus	Domains I,II,III of envelope protein	Immune	Fab, IgG	Chimpanzee	ELISA, immunoprecipitation, in vitro neutralization, in vivo protection	121
Japanese encephalitis virus	Envelope protein	Immune	Fab	Human	ELISA, immunoprecipitation, in vitro neutralization	247
Measles virus	Virus lysate	Immune (Puumala hantavirus!)	Fab	Human	ELISA, in vitro neutralization	248
Norovirus	Norovirus VLPs	Semisynthetic	scFv	Human	ELISA, immunoblot	249
Norwalk virus	Norwalk virus VLPs	Immune	Fab, IgG	Chimpanzee	ELISA, FACS, IF, in vitro inhibition	250
Poliovirus	Capsid proteins VP1 and VP3	Immune	Fab, IgG	Chimpanzee	ELISA, in vitro neutralization, in vivo protection	251
Puumala hantavirus	N protein, G2 protein	Immune	Fab	Human	ELISA	252
Puumala hantavirus	Gycoprotein G2	Immune	Fab	Human	ELISA, IF, in vitro neutralization	253
Rabies virus	Glycoprotein	Semisynthetic	scFv scFv‐Fc	Human	ELISA, immunoblot, in vitro neutralization	254
Rabies virus	Glycoprotein	Immune	scFv, IgG	Human	ELISA, flow cytometry, in vitro neutralization	153, 154
Rabies virus	n.d.	Immune	Fab	Human	ELISA	255
Rabies virus	Glycoprotein (antigenic site II)	Immune	Fab, IgG	Human	ELISA, immunostaining, Immunoblot, in vitro neutralization, in vivo protection	256
Rabies virus	Inactivated RABV	Naïve	VHH, VHH pentamer	Lama	ELISA, in vitro neutralization, in vivo protection	257
Rotavirus	NSP4	Semisynthetic	scFv	Human	ELISA, immunoblot	258
Rotavirus	VP8*	Semisynthetic	scFv	Human	ELISA, immunoblot, in vitro inhibition	259
Plum pox virus	NIa protease	Semisynthetic	scFv	Human	Immunoblot, dotblot	40
SARS‐CoV	S1 domain of spike protein	Naïve	scFv	Human	ELISA, in vitro neutralization	107
SARS‐CoV	S protein	Immune	Fab, IgG	Human	ELISA, IF, immuno blot, in vitro neutralization	260
SARS‐CoV	S protein	Immune	scFv	Chicken	ELISA, IF	261
Simian immunodeficiency virus	gp120 protein	Immune	Fab	Rhesus macaque	ELISA, in vitro neutralization	262
Vaccinia, variola virus	Vaccinia B5 envelope protein	Immune	Fab, IgG	Chimpanzee	ELISA, in vitro neutralization, in vivo protection	135
VEEV	Virus particles	Immune	scFv	Mice	ELISA	263
VEEV	E1/E2	Naïve	scFv, scFv‐Fc	Human	ELISA, immunoblot, IHC	105
VEEV	E1	Immune	scFv, scFv‐Fc	Macaque	ELISA, immunoblot, IHC, in vitro neutralization, in vivo protection	119
WEEV	n.d.	Immune	scFv, scFv‐Fc	Macaque	ELISA, IHC, in vitro neutralization	109
West Nile Virus	Domain I and II of WNV envelope protein	Naïve (nonimmune)	scFv	Human	ELISA, in vitro neutralization, in vivo protection	125
West Nile Virus	Domain IIII of WNV envelope protein	Immune	Fab	Human	ELISA, IF, in vitro neutralization, in vivo protection (failed)	264
White spot syndrome virus	Virus particles	Immune	scFv	Mice	ELISA, in vitro neutralization	265
Yellow fever virus	Domain II of envelope protein	Immune	scFv	Human	ELISA,immunoblot, immunoprecipitation, in vitro neutralization	113

ELISA, enzyme‐linked immunosorbent assays; HA, hemagglutinin; HCMV, human cytomegalovirus; HIV, human immunodeficiency virus; IF, immuno fluorescence microscopy; IHC, immuno histo chemistry; RABV, Rabies virus; SARS, severe acute respiratory syndrome; VLP, virus‐like particle; VEEV, Venezuelan equine encephalitis virus; WEEV, Western equine encephalitis virus.

This article is being made freely available through PubMed Central as part of the COVID-19 public health emergency response. It can be used for unrestricted research re-use and analysis in any form or by any means with acknowledgement of the original source, for the duration of the public health emergency.

4. Eukaryotic pathogens

Using phage display technology, a large panel of antibodies against a broad range of eukaryotic pathogens has been generated. These antibodies are directed against pathogenic animals, e.g. Taenia solium 157, protozoa, e.g. Cryptosporidium parvum 158, 159, Plasmodium falciparum 160, 161, or Toxoplasma gondii 162 and fungi such as Aspergillus fumigatus 41. Not only human pathogens have been in the focus of antibody generation, but also veterinary pathogens like Myxobolus rotundus 163 (a fish pathogen) or Babesia gibsoni (a dog pathogen) 164 and plant pathogens like Aspergillus niger 165, Fusarium verticilloides 166, or Sclerotinia sclerotiorum 167.

Most antibodies generated in these studies derive from human antibody gene libraries, but libraries from mouse 168, chicken 166, camel 169, or macaque 41 origin have been also successfully applied in phage display to generate monoclonal antibodies against eukaryotic pathogens.

In the following paragraphs detailed examples for the use of phage display for the generation of antibodies against different eukaryotic pathogens are given.

Aspergillus fumigates is the causative pathogen of allergic bronchopulmonary aspergillosis, chronic necrotizing aspergillosis, saprophytic aspergilloma, and highly lethal invasive aspergillosis, the most important Aspergillus‐related disease 170, 171. Invasive aspergillosis occurs in immunocompromised individuals for example after hematopoietic stem cell transplantation or solid organ transplantation 172. Due to its severity, an early diagnosis is crucial for a successful treatment. In 2009, Schütte et al. 41 reported the generation of several antibodies binding specifically to Crf2, an A. fumigates antigen of the glycosyl hydrolase family that is located in the cell wall of the growing hyphae. These antibodies could serve as tools for new diagnostic assays such as serum ELISA or histopathological immunofluorescence microscopy. The antibodies described in this study were isolated from two different phage‐displayed scFv libraries: One was a macaque immune library with a theoretical diversity of 1.5 × 10⁷ independent clones; while the other was human naïve antibody library HAL4/7 23. Both libraries were used in two different panning strategies: the first was performed on recombinant antigen immobilized on immunostrips; the second was carried out in solution with biotinylated antigen. Finally, 16 individual scFv clones were selected, with six clones deriving from the naïve library and ten from the immune library. Interestingly, all antibodies generated on immobilized antigen bind to linear epitopes, whereas all antibodies generated by panning in solution bind to conformational epitopes. Seven of these antibodies bound to their native antigen on growing hyphae of Aspergillus fumigatus and did not show cross‐reactions to other Aspergillus species or Candida albicans.

Malaria is a life‐threatening protozoal infection of the red blood cells and one of the most common mosquito‐borne diseases. Currently, an estimated 3.2 billion people in tropical and subtropical countries live at risk of malaria 173, 174. In humans malaria is elicited by at least five different species of Plasmodium, P. falciparum, P. vivax, P. ovale, P. malariae, and P. knowlesi 175. P. falciparum is responsible for most malaria‐related deaths globally, while P. vivax is the most widespread parasite 176. Three different approaches to generate antibodies against P. falciparum using phage display are described. In these studies various structures of different developmental stages of the parasite served as antigens. In a first study Roeffen et al. 160 used two scFv immune libraries, constructed from B‐lymphocytes from P. falciparum patients with transmission‐blocking immunity, to generate antibodies against Pfs48/45, a surface protein of P. falciparum that is expressed during zygote‐ and macrogamete stages. Pfs48/45 is a potential vaccine candidate since it is a target of transmission‐blocking antibodies that are taken up by the mosquito together with the blood meal and block within the mosquito's intestinal tract the oocyte development 177, 178, 179. The panning was performed on an extract of gametocytes immobilized in immunotubes and the elution was performed by competition with a mixture of four rat monoclonal mAbs, which recognize distinct epitopes on Pfs48/45 (epitopes I, IIb, III, and V). Interestingly, all selected 14 antibody clones bound to epitope III of Pfs48/45. In 2001, Sowa et al. 180 reported the isolation of a human monoclonal antibody against the Block 2 region of Plasmodium falciparum merozoite surface protein‐1 (PfMSP‐1) by phage display from a malaria patient derived scFv library. Lundquist et al. 161 generated a Fab‐immune library from leukocytes of 13 adults with acquired immunity to malaria. Three individual Fabs designated RAM1, RAM2, and RAM3 were isolated from this immune library by panning on a recombinant fragment of the merozite surface protein 3 (MSP‐3_194‐257). MSP‐3 is involved in heme binding, and antibodies against this protein promote eradication of the parasite by monocytes 181, 182. A synthetic peptide of the N‐terminal fragment of MSP‐3 is even used in human clinical trials as vaccine 183. The antibodies were subcloned into IgG1 and IgG3 format for further analysis. Binding of the antibodies to native parasite protein was demonstrated for all three antibodies in immuno blot and immunofluorescence microscopy. RAM1 and RAM2 also bound to their antigen in fixed and permeabilized cells in flow cytometry. The IgG3 format of RAM1 showed in a functional assay (antibody‐dependent cellular inhibition assay, ACDI) an inhibition rate that is comparable to affinity‐purified polyclonal anti‐MSP‐3_211–237 antibodies derived from immune donors. The IgG1 format also showed inhibition in the ACDI, but lower compared to IgG3 161.

An overview of recombinant antibodies generated by phage display against eukaryotic pathogens is given in Table 3.

Table 3.

Recombinant antibodies derived by phage display against eukaryotic pathogens

Eukaryotic pathogen	Target	Library type	Antibody format	Antibody origin	Application	Reference
Aspergillus fumigatus	Crf2	Immune/naïve	scFv, scFv‐Fc	Macaque/human	ELISA, IF	41
Aspergillus niger	Glucose oxidase	Semisynthetic	scFv	Human	ELISA	165
Babesia gibsoni	P50	Semisynthetic	scFv	Human	ELISA, IF	164
Candida albicans	Als3p and other	Naïve	scFv	Human	ELISA, IF, immunoblot, in vitro neutralization	266, 267, 268
Cryptosporidium parvum	P23	Semisynthetic	scFv	Human	ELISA	158
Cryptosporidium parvum	S16	Semisynthetic	scFv	Human	ELISA	159
Fusarium verticillioides	Cell wall‐bound proteins	Immune	scFv	Chicken	ELISA, IF	166
Fusarium verticillioides	Soluble cell wall‐bound proteins	Immune	scFv, scFv‐AP	Chicken	ELISA, IF, immunoblot	269
Myxobolus rotundus	Spore protein	Immune	scFv	Mice	ELISA, in vitro neutralization	163
Plasmodium chabaudi	AMA‐1	Immune	scFv	Mice	ELISA, immunoblot	168
Plasmodium falciparum	MSP‐1	Immune	scFv	Human	ELISA, IF	180
Plasmodium falciparum	MSP‐3	Immune	Fab, IgG	Human	ELISA, IF, immunoblot, FACS	161
Plasmodium falciparum	Pfs48/45	Immune	scFv	Human	ELISA, immunoblot	160
Plasmodium yoelii	MSP1	Immune	scFv	Mice	ELISA, immunoblot, in vivo protection	270
Sclerotinia sclerotiorum	SSPG1d	Immune	scFv	Mice	ELISA, immunoblot	167
Strongyloides venezuelensis	HSP60	Presumably naïve	scFv	Human	ELISA, IF	271
Taenia solium	TS14	Immune	VHH	Camel	ELISA, immunoblot	272
Taenia solium	T. solium metacestodes, peptides	Naïve	scFv	Human	ELISA, IF	157
Trypanosoma evansi	Different surface proteins	Immune	VHH	Camel	ELISA, FACS	169
Toxoplasma gondii	TgMIC2	Immune	scFv	Mice	ELISA, immunoblot	162

ELISA, enzyme‐linked immunosorbent assays; IHC, immuno histo chemistry; IF, immuno fluorescence microscopy; FACS, fluorescence‐activated cell sorting.

This article is being made freely available through PubMed Central as part of the COVID-19 public health emergency response. It can be used for unrestricted research re-use and analysis in any form or by any means with acknowledgement of the original source, for the duration of the public health emergency.

5. Antibodies against toxins

Several toxins are classified by the CDC as category A or B agents (for definition see: http://www.niaid.nih.gov/topics/biodefenserelated/biodefense/pages/cata.aspx) that are relevant for diagnostics and therapeutics. They can easily be disseminated and result in high or moderate mortality rates 184. In this context, the antibody phage display offers a powerful tool for antibody selection and allows the isolation of neutralizing antibodies against complete active toxins or special domains by using different human naïve antibody libraries with high diversity 185, 186, 187. For toxins, the aim is to find antibodies binding to the cell binding domain of the toxin and neutralize the interaction of this domain with the corresponding cellular target, mainly cell surface proteins. But also antibodies directed against the translocation domain or the enzymatic domain can neutralize the toxicitiy. For the isolation of high‐affinity antibodies against specific targets, animals are often immunized with toxoids, nontoxic subunits or selected toxin domains. Well suited are macaques for the construction of immune libraries, because macaque V‐genes are very similar to their human counterparts 59, 119, 188, 189, 190, 191.

In the following paragraphs, we give detailed examples for antibody generation using phage display and antibody engineering against different toxins.

So far, antibody phage display was successfully used for antibody selection against a panel of toxins classified as category A agents, such as from Clostridium botulinum (botulism) 190, 192, 193, 194 and Bacillus anthracis (anthrax) 59 and also against different category B agents, such as staphylococcal enterotoxin B 195, 196 and ricin toxin from Ricinus communis 191, 197. An example for a high‐risk microorganism that produces the most toxic substances known with the highest risk of potential use as bioweapons is the Gram‐positive, anaerobic, spore‐forming bacterium Clostridium botulinum, and other Clostridium spp. secreting eight different serotypes (A‐H) of botulinum neurotoxin (BoNT). Five serotypes (A, B, E, rarely F and only one case of H) are known to cause human botulism, a disease characterized by flaccid muscle paralysis requiring intensive hospital care and passive immunization 198, 199. Especially, serotype A is recognized as the most toxic substance known, showing LD₅₀ of about 1 ng/kg by intravenous route, about 10 ng/kg by the pulmonary route and about 1 μg/kg for the oral route 200. BoNTs are composed of a disulfide bond‐linked 50 kDa light chain and a 100 kDa heavy chain. The heavy chain contains two functional domains (Hc and Hn) that are responsible for toxin uptake into nerve cells by receptor‐mediated endocytosis and for the translocation of the light chain across the membrane into the neuronal cytosol. Whereas the catalytic domain of the light chain is responsible for the BoNT toxicity. The current approach for treatment of botulism includes the application of human anti‐botulism immunoglobulins, such as BabyBIG, or equine anti‐toxin serum. But the human serum stock of BabyBIG is limited and the equine anti‐toxin may cause hypersensitivity and serum sickness. In these situations, antibody phage display provides a technology to generate toxin‐neutralizing antibodies against each serotype. For instance, a macaque immune library was used to isolate neutralizing scFv with nM affinities against the light chain of BoNT/A 190, 201, 202, but also antibodies against the heavy chain or other relevant serotypes of BoNT are of therapeutic interest. Phage display technology was also used for isolation of single domain antibodies (VHH) after immunization of a llama with a cocktail of 7 BoNT toxoids (A‐F) 193. Another approach was the generation of a human antibody gene library after inducing a BoNT/A‐specific immune response by in vitro immunization 194. Furthermore, antibody phage display was used to generate antibodies against other clostridial toxins such as those from Clostridium tetani or Clostridium difficile 188, 203.

Anthrax, another serious infectious disease is caused by Bacillus anthracis, an aerobic, Gram‐positive, spore‐forming bacterium that is found in soils around the world. Bacillus anthracis secrets two toxins: the lethal toxin (LT) and the edema toxin (ET) 204. Both are composed of two subunits: the LT consists of the lethal factor (LF), and the protective antigen (PA); while the ET is formed by the edema factor, and PA. It was demonstrated that only LT has an essential role in the pathogenesis of anthrax 205. The subunit PA is the basis of current vaccines and induces the generation of neutralizing antibodies. In combination with antibiotics, commercial monoclonal antibodies against PA, such as raxibacumab, are commonly used for treatment 206. In 2012, the FDA approved raxibacumab to treat inhalational anthrax. Due to security issues the use of anti‐PA antibodies alone is questionable, since PA could be modified and lose the recognized epitopes while retaining biological activity. An alternative to anti‐PA antibodies are antibodies targeting the LF, such as 2LF, which was isolated from an immune library via antibody phage display technology 59. A combination of an anti‐PA antibody with an anti‐LF antibody could lead to a synergistic effect and improve the efficacy of the therapy.

An example for bacterial toxins classified as category B agent is staphylococcal enterotoxin B (SEB) from Staphylococcus aureus. This bacterium is a potential causative agent for food‐borne illness and produces twenty‐one types of staphylococcal enterotoxins that cause symptoms of food poisoning such as abdominal cramps, vomiting, and diarrhea 207, 208. SEB is a single polypeptide of 28 kDa and is the most potent toxin secreted by S. aureus. As a superantigen, it stimulates T cells and leads to an overproduction of cytokines, causing clinical symptoms such as fever, hypertension, and in some cases death. Phage display was used to generate recombinant antibodies from a murine immune library 196 and to identify the epitope of a SEB‐specific monoclonal antibody using a peptide phage library 209. Furthermore, a human monoclonal antibody against SEB was isolated from a synthetic human antibody gene library that inhibited SEB binding to MHC‐II 195.

The phage display technology was also used to isolate antibodies against ricin, which is also classified as category B agent by CDC. Ricin is a 61‐kDa glycoprotein from the castor bean plant (Ricinus communis), which consists of two distinct subunits (RTA and RTB). RTB is a galactose‐ and N‐acetylgalactosamine‐specific lectin that binds to specific sugar residues on the cell surface, allowing internalization of the toxic RTA by endocytosis 210. RTA is an RNA N‐glycosidase that irreversibly inactivates eukaryotic ribosomes, leading to the inhibition of protein synthesis 211. Human‐like antibodies were selected by phage display from a macaque immunized with RTA. One antibody, 43RCA, had a picomolar affinity and neutralized the biological activity of ricin in vitro 191. Furthermore, neutralizing VHH with high affinity was selected from a llama immune library. The best antibody C8 was able to neutralize 100% ricin activity in an in vitro assay using 40 μg/mL VHH 197.

In addition to the different toxins that are classified by the CDC as category A or B agents, the number of toxins is almost endless. Different animals are known to produce high potential toxins containing a complex composition. For example, Tityus serrulatus, known as Brazilian yellow scorpion and the most dangerous scorpion in Brazil, produces a 61‐amino acid peptide, called gamma toxin, which is the major toxic component in the venom 212, 213. Regarding this toxin, a neutralizing antibody was isolated from a human library via phage display and was capable of protecting mice 185. The same procedure was used for Bothrops jararacussu, a venomous pit viper species endemic in South America. By using a human antibody gene library, different antibodies were selected to inhibit the phospholipase activity of the venom in vitro and reduce the myotoxicity in vivo 214. Marine organisms can also produce toxins, e.g. the tetrodotoxin (TTX) from pufferfish. Here, scFv were selected from a human naïve antibody gene library neutralizing 99% of TTX activity in vitro 215.

An Overview of recombinant antibodies generated by phage display against toxins is given in Table 4.

Table 4.

Recombinant antibodies derived by phage display against toxins

Toxin	Species	Target	Library type	Antibody format	Antibody origin	Application	Reference
Androctonus australis venom	Androctonus australis hector	AahI' toxin	Immune	VHH, tandem VHH, VHH‐Fc	Camel	ELISA, in vivo protection	273
Anthrax toxin	Bacillus anthracis	Lethal factor (LF)	Immune	scFv	Macaque	ELISA, in vitro toxin neutralization, in vivo protection	59
Anthrax toxin	Bacillus anthracis	Protective antigen (PA)	Naïve	scFv	Human	In vitro toxin neutralization, in vivo protection	64
Bacillus thuringiensis toxin	Bacillus thuringiensis	Cry1C d‐endotoxins	Semisynthetic	scFv	Human	ELISA, in vitro toxin inhibition	274
Bee venom	Apis mellifera	Crude venom	Semisynthetic	scFv	Human	ELISA, in vitro toxin inhibition	275
Black widow venom	Latrodectus tredecimguttatus	Alpha‐latrotoxin	Immune	Fab	Mice	ELISA, in vitro inhibition, in vivo protection	276
Botulinum neurotoxin	Clostridium botulinum	Serotype A ‐ light chain	Immune	scFv	Macaque	ELISA, in vitro toxin inhibition	201
Botulinum neurotoxin	Clostridium botulinum	Serotype A ‐ light chain	Immune	scFv, scFv‐Fc	Macaque	ELISA, immunoblot, in vitro toxin inhibition, ex vivo toxin neutralization	190
Botulinum neurotoxin	Clostridium botulinum	Serotype A ‐ light chain	Immune	VHH	Camel	ELISA, immunoblot, in vitro neutralization	277
Botulinum neurotoxin	Clostridium botulinum	Serotype A ‐ heavy chain	Immune	scFv	Murine	ELISA, ex vivo toxin neutralization	278
Botulinum neurotoxin	Clostridium botulinum	Serotype A ‐ heavy chain	Immune	scFv	Human	ELISA, ex vivo toxin neutralization	192
Botulinum neurotoxin	Clostridium botulinum	Serotype A ‐ heavy chain	Naïve	scFv	Human	ELISA, ex vivo toxin neutralization	192
Botulinum neurotoxin	Clostridium botulinum	Serotype A ‐ heavy chain	Immune	scFv	Macaque	ELISA, ex vivo toxin neutralization	202
Botulinum neurotoxin	Clostridium botulinum	Serotype B ‐ light chain/heavy chain	Immune	scFv, scFv‐Fc	Macaque	ELISA, in vitro toxin inhibition, ex vivo toxin neutralization, in vivo protection	279
Botulinum neurotoxin	Clostridium botulinum	Serotype E ‐ light chain	Immune	scFv, scFv‐Fc	Macaque	ELISA, in vitro toxin inhibition, ex vivo toxin neutralization, in vivo protection	189
Botulinum neurotoxin	Clostridium botulinum	Serotype E ‐ heavy chain	Immune	VHH	Dromedary	ELISA, in vivo protection	280
Botulinum neurotoxin	Clostridium botulinum	Serotype A/B/C/D/E/F	Immune	VHH	Llama	ELISA, in vitro toxin inhibition	193
Brazilian yellow scorpion venom	Tityus serrulatus	Gamma‐toxin	Naïve	scFv	Human	ELISA, in vivo protection	185
C. difficile toxin	Clostridium difficile	TcdA	Immune	VHH	Llama	ELISA, immunoblot, in vitro neutralization	203
C. difficile toxin	Clostridium difficile	TcdA, TcdB	Immune	VHH, bispecific VHH	Alpaca	ELISA, in vitro toxin inhibition, in vivo protection	281
C. difficile toxin	Clostridium difficile	Binary CDT toxin	Immune	VHH, VHH‐Fc	Llama	ELISA, in vitro toxin inhibiton, IF	282
Enterotoxin B	Escherichia coli	EtxB	Semisynthetic	scFv	Human	ELISA, in vitro toxin inhibition	283
Fumonisin	Fusarium Verticillioides	B1	Semisynthetic	scFv	Human	—	284
Jararacussu venom	Bothrops jararacussu	Phospholipase A2 (PLA2)	Naïve	scFv	Human	ELISA, in vitro and in vivo protection	214
Mexican scorpion venom	Centruroides noxius	Cn2	Naïve	scFv	Human	ELISA, in vivo protection	285
Microcystin	Microcystis aeruginosa	ADDA	Semisynthetic	scFv	Human	ELISA	286
Pit Viper toxin	Trimeresurus mucros‐quamatus	VSP2	Immune	scFv	Chicken	ELISA, in vitro toxin neutralization, in vivo protection	287
Ricin	Ricinus communis	Chain A	Immune	scFv	Macaque	ELISA, in vitro toxin neutralization	191
Ricin	Ricinus communis	Chain A	Immune	VHH	Llama	ELISA, in vitro toxin neutralization	197
Shiga toxin	E. coli (STEC)	Stx2	Semisynthetic	Fab	Human	ELISA, immunoblot, in vitro neutralization	288
Shiga toxin	E. coli (EHEC)	Stx1, Stx2	Naïve	scFv	Human	ELISA, FACS, in vitro toxin neutralization	187
Shiga toxin	E. coli (STEC)	Stx1, Stx2	Immune	VHH	Alpaca	ELISA, in vitro toxin inhibition, in vivo protection	289
Staphylococcus enterotoxin B	Staphylococcus aureus	SEB	Immune	scFv	Mice	ELISA	196
Staphylococcus enterotoxin B	Staphylococcus aureus	SEB	Synthetic	Fab	Human	ELISA, immunoblot, in vitro toxin inhibition	195
Tetanus	Clostridium tetani	tetantus toxoid	Immune	Fab	Macaque	ELISA	188
Tetanus	Clostridium tetani	tetantus toxoid	Naïve	scFv	Human	ELISA, in vitro toxin inhibition	186
Tetrodotoxin (TTX)	Lagocephalus lunaris	C11H17N3O8	Naïve	scFv	Human	ELISA, in vitro and in vivo protection (prolonged survival)	215
Tetrodotoxin (TTX)	Lagocephalus lunaris	C11H17N3O8	Immune	scFv	Mice	ELISA	290
Thai cobra venom	Naja kaouthia	Neurotoxin	Naïve	scFv	Human	ELISA, immunoblot, in vivo protection	291
Vibrio parahaemolyticus hemolysin	Vibrio parahaemolyticus	TLH	Immune	scFv	Mice	ELISA, FACS, in vitro neutralization	292
Vibrio vulnificus toxin	Vibrio vulnificus	VvRtxA	Semisynthetic	scFv	Human	ELISA, in vitro toxin inhibition, IF	293

ELISA, enzyme‐linked immunosorbent assays.

This article is being made freely available through PubMed Central as part of the COVID-19 public health emergency response. It can be used for unrestricted research re-use and analysis in any form or by any means with acknowledgement of the original source, for the duration of the public health emergency.

6. Conclusion

Antibody phage display allows the generation of recombinant antibodies from different species, including human, llama, camel, macaque, shark, or mice. These antibodies are mainly derived from two types of sources: immune, or naïve libraries. Immune libraries should be preferred when immunized animals or convalescent patients are available, offering the chance to directly isolate affinity matured antibodies. If immunization is not possible or ethically not feasible, naïve antibody gene libraries are an alternative. In such an approach, the antibody generation process is not limited by the immune system. Using antibody phage display a variety of recombinant antibodies was generated for diagnostics and therapy against bacterial, viral pathogens, and eukaryotic pathogens as well as toxins.

The authors declare that there are no conflicts of interest.