Antibody light chain variable domains and their biophysically improved versions for human immunotherapy.

We set out to gain deeper insight into the potential of antibody light chain variable domains (VLs) as immunotherapeutics. To this end, we generated a naïve human VL phage display library and, by using a method previously shown to select for non-aggregating antibody heavy chain variable domains (VHs), we isolated a diversity of VL domains by panning the library against B cell super-antigen protein L. Eight domains representing different germline origins were shown to be non-aggregating at concentrations as high as 450 µM, indicating VL repertoires are a rich source of non-aggregating domains. In addition, the VLs demonstrated high expression yields in E. coli, protein L binding and high reversibility of thermal unfolding. A side-by-side comparison with a set of non-aggregating human VHs revealed that the VLs had similar overall profiles with respect to melting temperature (T(m)), reversibility of thermal unfolding and resistance to gastrointestinal proteases. Successful engineering of a non-canonical disulfide linkage in the core of VLs did not compromise the non-aggregation state or protein L binding properties. Furthermore, the introduced disulfide bond significantly increased their T(m)s, by 5.5-17.5 ° C, and pepsin resistance, although it somewhat reduced expression yields and subtly changed the structure of VLs. Human VLs and engineered versions may make suitable therapeutics due to their desirable biophysical features. The disulfide linkage-engineered VLs may be the preferred therapeutic format because of their higher stability, especially for oral therapy applications that necessitate high resistance to the stomach's acidic pH and pepsin.

Pentavalent single-domain antibodies reduce Campylobacter jejuni motility and colonization in chickens.

Campylobacter jejuni is the leading cause of bacterial foodborne illness in the world, with symptoms ranging from acute diarrhea to severe neurological disorders. Contaminated poultry meat is a major source of C. jejuni infection, and therefore, strategies to reduce this organism in poultry, are expected to reduce the incidence of Campylobacter-associated diseases. We have investigated whether oral administration of C. jejuni-specific single-domain antibodies would reduce bacterial colonization levels in chickens. Llama single-domain antibodies specific for C. jejuni were isolated from a phage display library generated from the heavy chain IgG variable domain repertoire of a llama immunized with C. jejuni flagella. Two flagella-specific single-domain antibodies were pentamerized to yield high avidity antibodies capable of multivalent binding to the target antigen. When administered orally to C. jejuni-infected two-day old chicks, the pentabodies significantly reduced C. jejuni colonization in the ceca. In vitro, the motility of the bacteria was also reduced in the presence of the flagella-specific pentabodies, suggesting the mechanism of action is through either direct interference with flagellar motility or antibody-mediated aggregation. Fluorescent microscopy and Western blot analyses revealed specific binding of the anti-flagella pentabodies to the C. jejuni flagellin.

Disulfide linkage engineering for improving biophysical properties of human VH domains.

To enhance their therapeutic potential, human antibody heavy chain variable domains (V(H)s) would benefit from increased thermostability. The highly conserved disulfide linkage that connects Cys23 and Cys104 residues in the core of V(H) domains is crucial to their stability and function. It has previously been shown that the introduction of a second disulfide linkage can increase the thermostability of camelid heavy-chain antibody variable domains (V(H)Hs). Using four model domains we demonstrate that this strategy is also applicable to human V(H) domains. The introduced disulfide linkage, formed between Cys54 and Cys78 residues, increased the thermostability of V(H)s by 14-18°C. In addition, using a novel hexa-histidine capture technology, circular dichroism, turbidity, size exclusion chromatography and multiangle light scattering measurements, we demonstrate reduced V(H) aggregation in domains with the Cys54-Cys78 disulfide linkage. However, we also found that the engineered disulfide linkage caused conformational changes, as indicated by reduced binding of the V(H)s to protein A. This indicates that it may be prudent to use the synthetic V(H) libraries harboring the engineered disulfide linkage before screening for affinity reagents. Such strategies may increase the number of thermostable binders.

Engineered single-domain antibodies with high protease resistance and thermal stability.

The extreme pH and protease-rich environment of the upper gastrointestinal tract is a major obstacle facing orally-administered protein therapeutics, including antibodies. Through protein engineering, several Clostridium difficile toxin A-specific heavy chain antibody variable domains (V(H)Hs) were expressed with an additional disulfide bond by introducing Ala/Gly54Cys and Ile78Cys mutations. Mutant antibodies were compared to their wild-type counterparts with respect to expression yield, non-aggregation status, affinity for toxin A, circular dichroism (CD) structural signatures, thermal stability, protease resistance, and toxin A-neutralizing capacity. The mutant V(H)Hs were found to be well expressed, although with lower yields compared to wild-type counterparts, were non-aggregating monomers, retained low nM affinity for toxin A, albeit the majority showed somewhat reduced affinity compared to wild-type counterparts, and were capable of in vitro toxin A neutralization in cell-based assays. Far-UV and near-UV CD spectroscopy consistently showed shifts in peak intensity and selective peak minima for wild-type and mutant V(H)H pairs; however, the overall CD profile remained very similar. A significant increase in the thermal unfolding midpoint temperature was observed for all mutants at both neutral and acidic pH. Digestion of the V(H)Hs with the major gastrointestinal proteases, at biologically relevant concentrations, revealed a significant increase in pepsin resistance for all mutants and an increase in chymotrypsin resistance for the majority of mutants. Mutant V(H)H trypsin resistance was similar to that of wild-type V(H)Hs, although the trypsin resistance of one V(H)H mutant was significantly reduced. Therefore, the introduction of a second disulfide bond in the hydrophobic core not only increases V(H)H thermal stability at neutral pH, as previously shown, but also represents a generic strategy to increase V(H)H stability at low pH and impart protease resistance, with only minor perturbations in target binding affinities. These are all desirable characteristics for the design of protein-based oral therapeutics.

Antibody recognition of a unique tumor-specific glycopeptide antigen.

Aberrant glycosylation and the overexpression of certain carbohydrate moieties is a consistent feature of cancers, and tumor-associated oligosaccharides are actively investigated as targets for immunotherapy. One of the most common aberrations in glycosylation patterns is the presentation of a single O-linked N-acetylgalactosamine on a threonine or serine residue known as the "Tn antigen." Whereas the ubiquitous nature of Tn antigens on cancers has made them a natural focus of vaccine research, such carbohydrate moieties are not always tumor-specific and have been observed on embryonic and nonmalignant adult tissue. Here we report the structural basis of binding of a complex of a monoclonal antibody (237mAb) with a truly tumor-specific glycopeptide containing the Tn antigen. In contrast to glycopeptide-specific antibodies in complex with simple peptides, 237mAb does not recognize a conformational epitope induced in the peptide by sugar substitution. Instead, 237mAb uses a pocket coded by germ-line genes to completely envelope the carbohydrate moiety itself while interacting with the peptide moiety in a shallow groove. Thus, 237mAb achieves its striking tumor specificity, with no observed physiological cross-reactivity to the unglycosylated peptide or the free glycan, by a combination of multiple weak but specific interactions to both the peptide and to the glycan portions of the antigen.

The role of CDR H3 in antibody recognition of a synthetic analog of a lipopolysaccharide antigen.

In order to explore the structural basis for adaptability in near germline monoclonal antibodies (mAb), we have examined the specificity of the promiscuous mAb S67-27 to both naturally derived carbohydrate antigens and a variety of synthetic nonnatural antigens based on the bacterial lipopolysaccharide component 3-deoxy-alpha-D-manno-oct-2-ulosonic acid (Kdo). One such analog, a 7-O-methyl (7-O-Me) Kdo disaccharide, was found to bind to the antibody with at least 30-fold higher affinity than any other antigen tested. The structure of S67-27 in complex with this analog and three other naturally occurring Kdo antigens revealed that the enhanced affinity of the mAb for the synthetic analog was accomplished by the strategic positioning of CDR H3 away from a conserved Kdo binding pocket that allowed the formation of new antibody-antigen contacts. Furthermore, the comparison of this structure with the structures of related mAbs revealed how the position and structure of CDR H3 influence the specificity or promiscuity of near-germline carbohydrate-recognizing antibodies by altering the architecture of the combining site.

Antibodies raised against chlamydial lipopolysaccharide antigens reveal convergence in germline gene usage and differential epitope recognition.

The structures of antigen-binding fragments from two related monoclonal antibodies have been determined to high resolution in the presence of several carbohydrate antigens raised against chlamydial lipopolysaccharide. With the exception of CDR H3, antibodies S54-10 and S73-2 are both derived from the same set of germline gene segments as the previously reported structures S25-2 and S45-18. Despite this similarity, the antibodies differ in specificity and the mechanism by which they recognize their cognate antigen. S54-10 uses an unrelated CDR H3 to recognize its antigen in a fashion analogous to S45-18; however, S73-2 recognizes the same antigen as S45-18 and S54-10 in a wholly unrelated manner. Together, these antibody-antigen structures provide snapshots into how the immune system uses the same set of inherited germline gene segments to generate multiple possible specificities that allow for differential recognition of epitopes and how unrelated CDR H3 sequences can result in convergent binding of clinically relevant bacterial antigens.

Pentabody-mediated antigen delivery induces antigen-specific mucosal immune response.

An efficient immunization system is essential for the development of mucosal vaccine. Cholera toxin (CT) and Escherichia coli heat labile toxin (LT) are among the strongest adjuvants tested in experimental animals but their use in humans has been hindered by their toxicity. On the other hand, the role of their non-toxic B-subunits, CTB or LTB, in enhancing mucosal immune response is not clear. We propose here a novel strategy for the induction of mucosal immune responses. Single domain antibodies (sdAbs) against a model antigen bovine serum albumin (BSA) were raised from the antibody repertoire of a llama immunized with BSA, pentamerized by fusing the sdAbs to CTB, generating the so-called pentabodies. These pentabodies were used to deliver the antigen by mixing the two components and administering the mixture to mice intranasally. One construct was equivalent to CT in helping induce mucosal immune response. It was also found that this ability was probably due to its high affinity to BSA, providing some insight into the controversial role of CTB in mucosal immunization: at least for BSA, the model antigen BSA employed in this study, CTB has to be tightly linked to the antigen to have adjuvant/immune-enhancing effect.

Exploration of specificity in germline monoclonal antibody recognition of a range of natural and synthetic epitopes.

To explore the molecular basis of antigen recognition by germline antibodies, we have determined to high resolution the structures of the near-germline monoclonal antibody S25-2 in complex with seven distinct carbohydrate antigens based on the bacterial sugar 3-deoxy-alpha-D-manno-oct-2-ulosonic acid (Kdo). In contrast to previous findings, the inherited germline Kdo monosaccharide binding site is not restricted to this bacterial sugar but is able to accommodate an array of substitutions and chemical modifications of Kdo, including naturally occurring antigens containing the related monosaccharide d-glycero-alpha-d-talo-oct-2-ulosonic acid as well as nonterminal Kdo residues. However, we show by surface plasmon resonance and ELISA how antibody S25-2 specificity is so dependent on the context in which the antigen is presented that a free disaccharide displays strong binding while the same lipid-A-bound disaccharide does not bind. These structures provide insight into how inherited germline genes code for immunoglobulins of limited flexibility that are capable of binding a range of epitopes from which affinity-matured antibodies are generated.

Recognition characteristics of monoclonal antibodies that are cross-reactive with gangliosides and lipooligosaccharide from Campylobacter jejuni strains associated with Guillain-Barré and Fisher syndromes.

The enteropathogen Campylobacter jejuni has the ability to synthesize glycan structures that are similar to mammalian gangliosides within the core component of its lipooligosaccharide (LOS). Exposure to ganglioside mimics in some individuals results in the production of autoantibodies that deleteriously attack nerve surface gangliosides, precipitating the onset of Guillain-Barré and Fisher syndromes (GBS and FS). We have characterized the interaction of four monoclonal antibodies (mAbs), established by sensitization of mice with LOS isolated from GBS- and FS-associated C. jejuni strains, with chemoenzymatically synthesized gangliooligosaccharides. Surface plasmon resonance (SPR) measurements demonstrate that three of the mAbs interact specifically with derivatives corresponding to their targeted gangliosides, with dissociation constants ranging from 10 to 20 microM. Antibody binding to the gangliooligosaccharides was probed by saturation transfer difference (STD) NMR spectroscopy. STD signals, resulting from antibody/oligosaccharide interaction, were observed for each of the four mAbs. In two cases, differential saturation transfer rates to oligosaccharide resonances enabled detailed epitope mapping. The binding of GD1a-S-Phe with GB1 is characterized by close association of the immunoglobulin with sites that are distributed over several residues of the oligosaccharide. This contrasts sharply with the profile observed for the binding of both GD3-S-Phe and GT1a-S-Phe with FS1. The close antigenic contacts in these ganglioside derivatives are confined to the N-acetylmannosaminyl portion of the terminal N-acetylneuraminic acid (NeuAc) residue of the disialosyl moiety. Our characterization of FS1 provides insight, at an atomic level, into how a single antigenic determinant presented by the LOS of C. jejuni can give rise to antibodies with binding promiscuity to [alphaNeuAc-(2-8)-alphaNeuAc]-bound epitopes and demonstrates why sera from FS patients have antibodies that are often reactive with more than one disialylated ganglioside.

The assembly of single domain antibodies into bispecific decavalent molecules.

Bispecific antibodies present unique opportunities in terms of new applications for engineered antibodies. However, designing ideal bispecific antibodies remains a challenge. Here we describe a novel bispecific antibody model in which five single domain antibodies (sdAbs) are fused via a linker sequence to the N-terminus of the verotoxin B (VTB) subunit, a pentamerization domain, and five sdAbs are fused via a linker sequence to the VTB C-terminus. Fifteen such decavalent bispecific molecules, termed decabodies, were constructed and characterized for the purpose of identifying an optimal decabody design. One of the fifteen molecules existed in a non-aggregated decavalent form. In conjunction with the isolation of sdAbs with the desired specificities from non-immune phage display libraries, the decabody strategy provides a means of generating high avidity bispecific antibody reagents, with good physical properties, relatively quickly.

A novel pentamer versus pentamer approach to generating neutralizers of verotoxin 1.

Verotoxins (VTs), or shiga-like toxins, are produced by enterohemorrhagic Escherichia coli (EHEC), which cause hemorrhagic colitis and hemolytic uremic syndrome. VTs are the major virulence factors in EHEC infection due to their cytotoxicity to various types of cells. Here, we present a novel type of VT neutralizer based on pentavalent single-domain antibodies, or pentabodies. Two single-domain antibodies (sdAbs) specific for the receptor binding sites of the B subunit of VT1 (VT1B) were isolated from a naïve llama phage display library. These two sdAbs were pentamerized to generate pentameric VT neutralizers, VTI-1 and VTI-3. Both VT neutralizers bound wild type VT1B specifically with superior functional affinity. In vitro neutralization assays showed that VTI-1 and VTI-3 were able to neutralize 90% and 40%, respectively, of the cytotoxicity caused by VT1. This effort provides the basis of a novel type of VT neutralizer that can potentially be produced at a relatively low cost.

Binding characteristics of a panel of monoclonal antibodies against the ligand binding domain of the human LDLr.

To obtain a panel of monoclonal antibodies (MAbs) to study the folding and conformation of the low density lipoprotein receptor (LDLr), we have generated hybridomas from LDLr-deficient mice that had been immunized with the extracellular domain of the human LDLr. The 12 MAbs were specific for the ligand binding domain of the LDLr, with individual MAbs recognizing epitopes in ligand binding repeats 1, 2, 3, 5, and 7. A subset of the MAbs failed to react with the LDLr when disulfide bonds were reduced, and one MAb, specific for an epitope that spans ligand binding repeats 1 and 2, recognized two conformational forms of the LDLr with different affinities. Antibodies specific for ligand binding repeats 3, 5, and 7 completely blocked the binding of LDL particles to the LDLr on cultured human fibroblasts, whereas MAbs with epitopes in ligand binding repeats 1 and 2 partially blocked the binding of LDL to the LDLr. These anti-LDLr MAbs will serve as useful probes for further analysis of LDLr conformation and LDLr-mediated lipoprotein binding.

Selection, characterization, and CDR shuffling of naive llama single-domain antibodies selected against auxin and their cross-reactivity with auxinic herbicides from four chemical families.

Indoleacetic acid (IAA)-binding single-domain antibodies (sdAbs) were isolated from a naive phage-display library constructed from the heavy chain antibody repertoire of a Ilama. The highest-affinity sdAb isolated (CSF2A) had a K(D) of 5-20 microM for two IAA-protein conjugates and a K(D) of 20 microM for free IAA. This sdAb also bound to a synthetic auxin analogue, 1-naphthaleneacetic acid (NAA), and to six auxinic herbicides (K(D) values of 0.5-2 mM), but not to serotonin and tryptophan, which are structurally similar to IAA but have no auxinic activity. To understand how sdAb CSF2A binds IAA and to determine which complementary-determining region(s) (CDR) participate(s) most in binding IAA, CSF2A was shuffled with four other sdAb clones by staggered extension process (StEP). After panning against IAA, two shuffled sdAbs were found: sdAb CSB1A, which originated from three different parental clones, and sdAb CSE8A, derived from two parental clones. These shuffled sdAbs and CSF2A were each fused to the B subunit of the Escherichia coli verotoxin, resulting in the formation of the pentamerized sdAbs V2NCSB1A, V2NCSE8A, and V2NCSF2A, which were analyzed by surface plasmon resonance (SPR) along with the sdAbs previously isolated. The shuffled clones had affinity for IAA (20 microM) similar to that of the highest affinity parental clone CSF2A, but much lower affinity for the auxinic herbicides. CDR2 was instrumental in binding IAA, whereas hydrophobic CDR3 was important for binding the auxinic herbicides. A novel SPR methodology is also described for specific immobilization of pentamerized sdAbs, allowing determination of K(D) values of Ab interaction with underivatized, low molecular weight haptens.

Characterization of murine monoclonal antibodies against Helicobacter pylori lipopolysaccharide specific for Lex and Ley blood group determinants.

The cell envelope of Helicobacter pylori contains lipopolysaccharide (LPS), the O-chain of which expresses type 2 Lex and Ley blood group antigens, which mimic human gastric mucosal cell-surface glycoconjugates and may contribute to the survival of H. pylori in gastric mucosa. Here we describe the generation of monoclonal antibodies specific for Lex and Ley blood group determinants and the characterization of their binding properties using purified, structurally defined H. pylori LPS, synthetic glycoconjugates, and H. pylori cells. Analysis of oligosaccharide binding by SPR provided a rapid and reliable means for characterization of antibody affinities. One of the antibodies, anti-Lex, was of IgG3 subclass and had superior binding characteristics as compared with the commercially available anti-Lex IgM. These antibodies could have potential in the immunodiagnosis of certain types of cancer, in serotyping of H. pylori isolates, and in structure-function studies.

Isolation of monomeric human V(H)s by a phage selection.

Human V(H) domains are promising molecules in applications involving antibodies, in particular, immunotherapy because of their human origin. However, they are, in general, prone to aggregation. Therefore, various strategies have been employed to acquire monomeric human V(H)s. We had previously discovered that filamentous phages displaying engineered monomeric V(H) domains gave rise to significantly larger plaques on bacterial lawns than phages displaying wild type V(H)s with aggregation tendencies. Using plaque size as the selection criterion and a phage-displayed naïve human V(H) library we identified 15 V(H)s that were monomeric. Additionally, the V(H)s demonstrated good expression yields, good refolding properties following thermal denaturation, resistance to aggregation during long incubation at 37 degrees C, and to trypsin at 37 degrees C. These 15 V(H)s should serve as good scaffolds for developing immunotherapeutics, and the selection method employed here should have general utility for isolating proteins with desirable biophysical properties.

Transgenic tobacco plants expressing a dimeric single-chain variable fragment (scfv) antibody against Salmonella enterica serotype Paratyphi B.

Transgenic tobacco plants were produced that express an anti-Salmonella enterica single-chain variable fragment (scFv) antibody that binds to the lipopolysaccharide (LPS) of S. enterica Paratyphi B. The coding sequence of this scFv was optimized for expression in tobacco, synthesized and subsequently placed behind three different promoters: an enhanced tobacco constitutive ubiquitous promoter (EntCUP4), and single- and double-enhancer versions of the Cauliflower Mosaic Virus 35S promoter (CaMV 35S). These chimeric genes were introduced into Nicotiana tabacum cv. 81V9 by Agrobacterium-mediated transformation and 50 primary transgenic (T(0)) plants per construct were produced. Among these plants, 23 were selected for the ability to express active scFv as determined by enzyme-linked immunosorbent assay (ELISA) using S. enterica LPS as antigen. Expanded bed adsorption-immobilized metal affinity chromatography (EBA-IMAC) was used to purify 41.7 mug of scFv/g from leaf tissue. Gel filtration and surface plasmon resonance (SPR) analyses demonstrated that the purified scFv was active as a dimer or higher-order multimer. In order to identify T(1) plants suitable for development of homozygous lines with heritable scFv expression, kanamycin-resistance segregation analyses were performed to determine the number of T-DNA loci in each T(0) plant, and quantitative ELISA and immunoblot analyses were used to compare expression of active and total anti-Salmonella scFv, respectively, in the T(1) generation. As S. enterica causes millions of enteric fevers and hundreds of thousands of deaths worldwide each year, large-scale production and purification of this scFv will have potential for uses in diagnosis and detection, as a therapeutic agent, and in applications such as water system purification.

Affinity maturation of a V(H)H by mutational hotspot randomization.

V(H)Hs from naive libraries have dissociation constants (K(D)s) in the low micromolar range and thus, for most antibody applications, their intrinsic affinities need to be improved significantly. Non-targeted in vitro affinity maturation approaches based on indiscriminate randomization of complementarity-determining region (CDR) residues or random mutagenesis of conventional antibody variable domains have been shown to improve the affinity of recombinant antibodies by 450- to over 6000-fold. A different, targeted approach based on selective randomization of CDR codons containing AGY/RGYW nucleotide mutational hotspots i.e., "hotspot codons", also promises to be very efficient for improving antibody affinities. Here we employed the latter approach for improving the affinity of PTH22, a parathyroid hormone (PTH)-derived peptide-specific V(H)H that was isolated from a naive llama phage display library. A PTH22 mutant ribosome display library was constructed by randomizing nine CDR2 and CDR3 hotspot codons. The affinity improvement of the lead binder was 30-fold, which seems somewhat low in view of the large number of randomized hotspot codons. Nucleotide sequence analyses of PTH22 and 23 naive V(H)Hs suggested that many AGY/RGYW mutational hotspots are not affinity mutational hotspots but play a role in V(H)H solubility, structure, and deletion/insertion events. Our results indicate that the mutagenesis approach described here is beneficial in terms of yielding moderate increases in affinity while fine-tuning physical properties of an antibody.

X-ray crystal structure of a galactose-specific C-type lectin possessing a novel decameric quaternary structure.

Rattlesnake venom lectin (RSL) from the western diamondback rattlesnake (Crotalus atrox) is an oligomeric galactose-specific C-type lectin. The X-ray crystal structure of RSL, in complex with lactose and thiodigalactoside, at 2.2 and 2.3 A resolution, respectively, reveals a decameric protein composed of two 5-fold symmetric pentamers arranged in a staggered, back-to-back orientation. Each monomer corresponds to a single canonical C-type lectin carbohydrate recognition domain devoid of accessory domains and is disulfide-bonded to a monomer in the other pentamer. The structure is the first example of that of a carbohydrate complex of a vertebrate galactose-specific C-type lectin. The 10 carbohydrate-binding sites, located on the rim of the decamer, suggest a role for multivalent interactions and a mechanism for RSL's ability to promote receptor cross-linking and cell aggregation.

Pentamerization of single-domain antibodies from phage libraries: a novel strategy for the rapid generation of high-avidity antibody reagents.

We describe a novel type of molecule in which single-domain antibodies (sdAbs) isolated from a nai;ve llama single domain antibody library are linked to an oligomerization domain to generate high-avidity, antigen-binding reagents. An sdAb is fused to the B-subunit of Escherichia coli verotoxin, or shiga-like toxin, which self-assembles to form a homopentamer and results in simultaneous sdAb pentamerization and introduction of avidity. Molecular modeling indicated that this fusion protein (PDB: 1OJF), termed pentabody, has structural flexibility for binding to surface-presented antigen. In the instance of an sdAb specific for a peptide antigen, pentamerization resulted in a dramatic increase in functional affinity for immobilized antigen. The pentabody was expressed in high yield in E.coli in a non-aggregated state, and exhibited excellent thermostability and protease resistance. This technology provides a relatively rapid means of generating novel antigen-binding molecules that bind strongly to immobilized antigen. It is expected that pentavalent sdAbs will have general applicability in proteomics, immunochemical staining, cancer diagnosis and other applications in which antigens are presented multivalently.