Dimeric and trimeric antibodies: high avidity scFvs for cancer targeting

Recombinant antibody fragments can be engineered to assemble into stable multimeric oligomers of high binding avidity and specificity to a wide range of target antigens and haptens. This review describes the design and expression of diabodies (dimers), triabodies (trimers) and tetrabodies (tetramers). In particular we discuss the role of linker length between V-domains and the orientation of the V-domains to direct the formation of either diabodies (60 kDa), triabodies (90 kDa) or tetrabodies (120 kDa), and how the size, flexibility and valency of each molecules is suited to different applications for in vivo imaging and therapy. Single chain Fv antibody fragments joined by polypeptide linkers of at least 12 residues irrespective of V-domains orientation predominantly form monomers with varying amounts of dimer and higher molecular mass oligomers in equilibrium. A scFv molecule with a linker of 3-12 residues cannot fold into a functional Fv domain and instead associates with a second scFv molecule to form a bivalent dimer (diabody, approximately 60 kDa). Reducing the linker length below three residues can force scFv association into trimers (triabodies, approximately 90 kDa) or tetramers ( approximately 120 kDa) depending on linker length, composition and V-domain orientation. A particular advantage for tumour targeting is that molecules of 60-100 kDa have increased tumour penetration and fast clearance rates compared with the parent Ig (150 kDa). We highlight a number of cancer-targeting scFv diabodies that have undergone successful pre-clinical trials for in vivo stability and efficacy. We also briefly review the design of multi-specific Fv modules suited to cross-link two or more different target antigens. Bi-specific diabodies formed by association of different scFv molecules have been designed as cross-linking reagents for T-cell recruitment into tumours (immunotherapy), viral retargeting (gene therapy) and as red blood cell agglutination reagents (immunodiagnostics). The more challenging trispecific multimers (triabodies) remain to be described.

Ligand-binding characteristics of recombinant amino- and carboxyl-terminal fragments of human insulin-like growth factor-binding protein-3

Insulin-like growth factor-binding protein-3 (IGFBP-3) is a member of a family of structurally conserved proteins (IGFBP-1 to -6) which act as carriers and regulators of the mitogenic peptide hormones IGF-I and IGF-II. Members of the IGFBP family share conserved cysteine-rich amino- and carboxyl-terminal regions. The amino-terminal domain of these proteins is recognised to contain an IGF-binding determinant, but evidence to support a binding site in the carboxyl-terminal region of the protein is less rigorous. To further investigate this, we have synthesised both the amino-terminal (residues 1-88; N-88) and carboxyl-terminal (residues 165-264; C-165) domains of human IGFBP-3 in bacteria, as fusion proteins with a carboxyl-terminal FLAG peptide. Although only C-165 showed binding to IGF-I and -II by solution-binding assays, both N-88 and C-165 demonstrated binding to IGF-I and -II by biosensor analysis albeit with reduced affinities compared with full-length IGFBP-3. Only the carboxyl-terminal fragment (C-165) was able to form hetero-trimeric complexes with IGF-I and the acid-labile subunit (ALS). We conclude that the carboxyl-terminal domain of IGFBP-3 contains an IGF-binding determinant and can form ternary complexes with ALS.

Design and application of diabodies, triabodies and tetrabodies for cancer targeting

Multivalent recombinant antibody fragments provide high binding avidity and unique specificity to a wide range of target antigens and haptens. This review describes the design and expression of diabodies, triabodies and tetrabodies using examples of scFv molecules that target viruses (influenza neuraminidase) and cancer (Ep-CAM; epithelial cell adhesion molecule). We discuss the preferred choice of linker length between V-domains to direct the formation of either diabodies (60 kDa), triabodies (90 kDa) or tetrabodies (120 kDa), each with size, flexibility and valency suited to different applications for in vivo imaging and therapy. The increased binding valency of these scFv multimers results in high avidity (low off-rates). A particular advantage for tumour targeting is that molecules of 60-100 kDa have increased tumour penetration and fast clearance rates compared to the parent Ig (150 kDa). We highlight a number of cancer-targeting scFv multimers that have recently successfully undergone pre-clinical trials for in vivo stability and efficacy. We also review the design of multi-specific Fv modules suited to cross-link two or more different target antigens. These bi- and tri-specific multimers can be formed by association of different scFv molecules and, in the first examples, have been designed as cross-linking reagents for T-cell recruitment into tumours (immunotherapy), viral retargeting (gene therapy) and as red blood cell agglutination reagents (immunodiagnostics).

Expression and purification of antigenically active soluble derivatives of the heterodimeric and homodimeric forms of the mouse CD8 lymphocyte membrane glycoprotein

The T lymphocyte membrane glycoprotein CD8 enhances antigen recognition by class I-restricted T cells. There are two naturally occurring forms of CD8, an alphabeta heterodimer expressed by the majority of CD8(+) T cells, and a less abundant alphaalpha homodimer found on specialised T cell subsets. An expression strategy was developed for production of soluble CD8alphaalpha and CD8alphabeta extracellular domains for use in ligand binding studies. Mouse CD8alpha was expressed autonomously as a homodimer at 10 mg/l in mammalian fibroblasts, but CD8beta was not expressed at significant levels in the absence of CD8alpha. Co-expression with CD8alpha led to significant enhancement in the level of CD8beta expression, which was secreted as a non-covalent heterodimer at 3 mg/l with CD8alpha. Despite the marked increase of CD8beta expression in the presence of CD8alpha, an excess of soluble CD8alphaalpha homodimer was also present in the supernatant of co-expressing cell clones. In order to resolve the CD8alphaalpha homodimer from the CD8alphabeta heterodimer, affinity chromatographic techniques specific for the CD8beta subunit were employed. Purification procedures requiring elution from affinity matrices at low pH led to substantial losses in the total antigenic activity and partial subunit dissociation of the soluble CD8alphabeta heterodimer. The inclusion of a hexahistidine tag at the C-terminus of CD8beta enabled affinity purification of soluble CD8alphabeta (and sCD8alphaalpha) under neutral conditions, yielding recombinant protein with the correct stoichiometry and full antigenic activity. This method may prove useful for production of other soluble recombinant heterodimeric receptor proteins whose antigenicity is affected by denaturation during immunoaffinity purification.

ScFv multimers of the anti-neuraminidase antibody NC10: shortening of the linker in single-chain Fv fragment assembled in V(L) to V(H) orientation drives the formation of dimers, trimers, tetramers and higher molecular mass multimers

Synthetic genes encoding single-chain variable fragments (scFvs) of NC10 anti-neuraminidase antibody were constructed by joining the V(L) and V(H) domains with linkers of fifteen, five, four, three, two, one and zero residues. These V(L)-V(H) constructs were expressed in Escherichia coli and the resulting proteins were characterized and compared with the previously characterized NC10 scFv proteins assembled in V(H)-V(L) orientation. Size-exclusion chromatography and electron microscope images of complexes formed between various NC10 scFvs and anti-idiotype Fab' were used to analyse the oligomeric status of these scFvs. The result showed that as the linker length between V(L) and V(H) was reduced, different patterns of oligomerization were observed compared with those with V(H)-V(L) isomers. As was the case for V(H)-V(L) orientation, the scFv-15 V(L)-V(H) protein existed mainly as a monomer whereas dimer (diabody) was a predominant conformation for the scFv-5, scFv-4 and scFv-3 V(L)-V(H) proteins. In contrast to the V(H)-V(L) isomer, direct ligation of V(L) to V(H) led to the formation of predominantly a tetramer (tetrabody) rather than to an expected trimer (triabody). Furthermore, the transition between dimers and higher order oligomers was not as distinct as for V(H)-V(L). Thus reducing the linker length in V(L)-V(H) from three to two residues did not precisely dictate a transition between dimers and tetramers. Instead, two-residue as well as one-residue linked scFvs formed a mixture of dimers, trimers and tetramers.

High avidity scFv multimers; diabodies and triabodies

Multivalent recombinant antibody fragments provide high binding avidity and unique specificity to a wide range of target antigens and haptens. This review describes how careful choice of linker length between V-domains creates new types of Fv modules with size, flexibility and valency suited to in vivo imaging and therapy. Further, we review the design of multi-specific Fv modules suited to cross-linking target antigens for cell-recruitment, viral delivery and immunodiagnostics. Single chain Fv antibody fragments (scFvs) are predominantly monomeric when the V(H) and V(L) domains are joined by polypeptide linkers of at least 12 residues. An scFv molecule with a linker of 3 to 12 residues cannot fold into a functional Fv domain and instead associates with a second scFv molecule to form a bivalent dimer (diabody, approximately 60 kDa). Reducing the linker length below three residues can force scFv association into trimers (triabodies, approximately 90 kDa) or tetramers ( approximately 120 KDa) depending on linker length, composition and V-domain orientation. The increased binding valency in these scFv multimers results in high avidity (long off-rates). A particular advantage for tumor targeting is that molecules of approximately 60-100 kDa have increased tumor penetration and fast clearance rates compared to the parent Ig. A number of cancer-targeting scFv multimers have recently undergone pre-clinical evaluation for in vivo stability and efficacy. Bi- and tri-specific multimers can be formed by association of different scFv molecules and, in the first examples, have been designed as cross-linking reagents for T-cell recruitment into tumors (immunotherapy) and as red blood cell agglutination reagents (immunodiagnostics).

Utilization of the Streptoalloteichus hindustanus resistance determinant ShBle as a protein framework: effect of mutation upon ShBle dimerization and interaction of C-terminal displayed peptide epitopes

We have selected the Streptoalloteichus hindustanus bleomycin-resistance protein ShBle, a 28-kDa homodimer, as a scaffold for the display of bioactive peptides and other peptide epitopes. To create a monomeric scaffold, we investigated the effect of mutating residue proline 9 to glycine. This residue plays a critical role in ShBle dimerization by affecting the position of the eight N-terminal residues which secure the interaction between the monomeric subunits. We demonstrate that this mutation weakens the dimerization interaction, resulting in establishment of a stable equilibrium between monomeric and dimeric ShBle species in solution. Circular dichroism and SDS-PAGE data indicate that the Pro9Gly mutation does not disrupt the structure of the molecule. Production of a fully monomeric form of ShBle required complete removal of the eight-residue N-terminal peptide, and the interaction across the now solvent-exposed hydrophobic interface of the ShBle monomer was insufficient to drive dimerization. To demonstrate efficient display of epitope tags on the ShBle protein, we displayed dual-octapeptide FLAG tags at the protein C-terminus. These additions did not interfere with protein folding or activity. The resulting ShBle scaffold was used to compare the efficiency of two commercial FLAG-specific antibodies by biosensor.

Analysis of the binding of the Fab fragment of monoclonal antibody NC10 to influenza virus N9 neuraminidase from tern and whale using the BIAcore biosensor: effect of immobilization level and flow rate on kinetic analysis

The binding of the Fab fragment of monoclonal antibody NC10 to influenza virus N9 neuraminidase, isolated from tern and whale, was measured using an optical biosensor. Both neuraminidases, homotetramers of 190 kDa, were immobilized to avoid multivalent binding, and the binding of the monovalent NC10 Fab to immobilized neuraminidase was analyzed using the 1:1 Langmuir binding model. A contribution of mass transport to the kinetic constants was demonstrated at higher surface densities and low flow rates, and was minimized at low ligand densities and relatively high flow rates (up to 100 microl/min). Application of a global fitting algorithm to a 1:1 binding model incorporating a correction term for mass transport indicated that mass transport was minimized under appropriate experimental conditions; analysis of binding data with a mass transport component, using this model, yielded kinetic constants similar to those obtained with the 1:1 Langmuir binding model applied to binding data where mass transport had been minimized experimentally. The binding constant for binding of NC10 Fab to N9 neuraminidase from tern influenza virus (K(A) = 6.3 +/- 1.3 x 10(7) M(-1)) was about 15-fold higher than that for the NC10 Fab binding to N9 neuraminidase from whale influenza virus (K(A) = 4.3 +/- 0.7 x 10(6) M(-1)). This difference in binding affinity was mainly attributable to a 12-fold faster dissociation rate constant of the whale neuraminidase-NC10 Fab complex and may be due to either (i) the long-range structural effects caused by mutation of two residues distant from the binding epitope or (ii) differences in carbohydrate residues, attached to Asn(200), which form part of the binding epitope on both neuraminidases to which NC10 Fab binds.

scFv multimers of the anti-neuraminidase antibody NC10: length of the linker between VH and VL domains dictates precisely the transition between diabodies and triabodies

Single-chain Fv antibody fragments (scFvs) incorporate a polypeptide linker to tether the VH and VL domains together. An scFv molecule with a linker 5-12 residues long cannot fold into a functional Fv domain and instead associates with a second scFv molecule to form a bivalent dimer (diabody). Direct ligation of VH and VL domains further restricts association and forces three scFv molecules to associate into a trivalent trimer (triabody). We have defined the effect of linker length on scFv association by constructing a series of scFvs from anti-neuraminidase antibody NC10 in which the linker varied from one to four glycine residues. NC10 scFv molecules containing linkers of three and four residues showed a strong preference for dimer formation (diabodies), whereas a linker length of one or two glycine residues prevented the formation of diabodies and directed scFv association into trimers (triabodies). The data suggest a relatively strict transition from dimer (diabody) to trimer (triabody) upon reduction of the linker length from three to two glycine residues. Modelling studies are consistent with three residues as the minimum linker length compatible with diabody formation. Electron microscope images of complexes formed between the NC10 scFv multimers and an anti-idiotype Fab' showed that the dimer was bivalent for antigen binding and the trimer was trivalent.

Determination of binding constants by equilibrium titration with circulating sample in a surface plasmon resonance biosensor

A commercial surface plasmon resonance biosensor, BIACORE X, is employed as a detector in a closed loop of a small sample volume. The sample is continuously circulated by an external syringe pump over two sensor spots, one functionalized with immobilized binding sites to a soluble binding partner in the mobile phase and one serving as a reference surface. A binding isotherm for the interacting macromolecules can be obtained by a stepwise titration of the soluble reactant into the circulating loop, each step followed by observation of the signal increase until equilibrium is attained. Binding constants can be measured under conditions free of mass transport artifacts and without the requirement for regeneration of the immobilized binding sites. This procedure is similar to the stepwise titration procedure described for the cuvette-based sensor design (D. R. Hall and D. J. Winzor, 1997, Anal. Biochem. 244, 152-160). In the presented configuration, the high baseline stability of the instrument combined with the availability of a reference surface for the detection of nonspecific binding permits refractive index changes upon addition of the aliquots to be measured, as well as accounting for temperature or instrumental drifts, and allows for a very long experimental time. This feature extends the applicability of equilibrium titration to systems with higher affinity or slower dissociation rate constants. Furthermore a solution competition titration is described that avoids artifacts from the immobilization procedure to provide a method for measurement of binding constants in solution. Kinetic information on the complex dissociation can also be obtained by combination of sample delivery via the external pump with the injection of competitor via the microfluidics of the biosensor. The rapid injection of high concentrations of competitor allows the observation of fast dissociation processes under conditions minimizing rebinding.

Three-dimensional structures of single-chain Fv-neuraminidase complexes

The structure of the complex between a recombinant single-chain Fv construct of antibody NC10 with a five-residue peptide linker between VH and VL (termed scFv(5)), and its antigen, tetrameric neuraminidase from influenza virus (NA), has been determined and refined at 2.5 A resolution. The antibody-antigen binding interface is very similar to that of a similar NC10 scFv-NA complex in which the scFv has a 15-residue peptide linker (scFv(15)), and the NC10 Fab-NA complex. However, scFv(5) and scFv(15) have different stoichiometries in solution. While scFv(15) is predominantly monomeric in solution, scFv(5) forms dimers exclusively, because the five-residue linker is not long enough to permit VH and VL domains from the same polypeptide associating and forming an antigen-binding site. Upon forming a complex with NA, scFv(15) forms a approximately 300 kDa complex corresponding to one NA tetramer binding four scFv(15) monomers, while scFv(5) forms a approximately 590 kDa complex, corresponding to two NA tetramers crosslinked by four bivalent scFv(5) dimers. However, the dimeric scFv(5) in the scFv(5)-NA crystals does not crosslink NA tetramers, and modelling studies indicate that it is not possible to pack four dimeric and simultaneously bivalent scFvs between the NA tetramers with only a five-residue linker between VH and VL. The inability arises from the exacting requirement to orient the two antigen-binding surfaces to bind the tetrameric NA antigen while avoiding steric clashes with NC10 scFv(5) dimers bound to other sites on the NA tetramer. The utility of bivalent or bifunctional scFvs with short linkers may therefore be restricted by the steric constraints imposed by binding multivalent antigens.

Orientation of antigen binding sites in dimeric and trimeric single chain Fv antibody fragments

Electron microscopy of dimeric and trimeric single chain antibody Fv fragments (scFvs) complexed with anti-idiotype Fab fragments was used to reveal the orientation of antigen binding sites. This is the first structural analysis that discloses the multivalent binding orientation of scFv trimers (triabodies). Three different scFv molecules were used for the imaging analysis; NC10 scFv-5 and scFv-0, with five- and zero-residue linkers respectively between the VH and VL domains, were complexed with 3-2G12 anti-idiotype Fab fragments and 11-1G10 scFv-0 was complexed with NC41 anti-idiotype Fab fragments. The scFv-5 molecules formed bivalent dimers (diabodies) and the zero-linker scFv-0 molecules formed trivalent trimers (triabodies). The images of the NC10 diabody-Fab complex appear as boomerangs, not as a linear molecule, with a variable angle between the two Fab arms and the triabody-Fab complexes appear as tripods.

Single-chain Fv of anti-idiotype 11-1G10 antibody interacts with antibody NC41 single-chain Fv with a higher affinity than the affinity for the interaction of the parent Fab fragments

A single-chain Fv (scFv) fragment of anti-idiotype antibody 11-1G10, which recognizes an idiotope of anti-neuraminidase antibody NC41, was constructed by joining VH and VL domains with a (Gly4Ser)3 linker, with a pelB leader sequence, and two C-terminal FLAG tag sequences, and expressed in E. coli (10 mg/L). The 11-1G10 scFv was isolated by affinity chromatography on an anti-FLAG M2 antibody column as a 2:1 mixture of monomer and dimer forms which were separated by Superdex 75 chromatography; monomer (at 100 microg/ml) was stable for 7 days at 21 degrees C and 30 days at 4 degrees C, whereas the dimer slowly dissociated to monomer to yield a 2:1 monomerdimer equilibrium mixture after 30 days at 4 degrees C. The dimer was bivalent, with each combining site binding an NC41 Fab to yield a stable complex of Mr approximately 156,000. Binding affinities, determined in solution using a BIAcore biosensor, showed that the affinity for the interaction of 11-IG10 scFv monomer with NC41 scFv monomer was five- to six-fold higher than the interaction of the parent Fab pair. This is the first example of an scFv derived from a monoclonal antibody with a higher affinity than its parent Fab.

Effects of substitutions in the binding surface of an antibody on antigen affinity

The interactions between the Fab and single-chain Fv (scFv) fragments of an antibody (NC10) and its antigen, influenza virus neuraminidase, were analysed in the crystal structures of the Fab-neuraminidase and scFv-neuraminidase complexes. To investigate the contribution to binding made by cavities, salt links and hydrogen bonds in the antibody-antigen interface, 14 single amino acid replacements were made at six contact residues in the scFv fragment by site-directed mutagenesis. The binding affinity of each mutant scFv antibody for neuraminidase was determined with a BIAcore optical biosensor. Four of the mutations resulted in large changes in the free energy of binding to neuraminidase (deltadeltaG > 1 kcal/mol) and together may account for approximately 70% of the free energy of binding. Hence these data support the theory that a small number of residues form the 'functional epitope' and are most important for binding of NC10 to neuraminidase. The salt link between antibody residue (Asp)H56 and (Lys)N432 from neuraminidase was demonstrated to be important for affinity, since substitution of (Asp)H56 with Asn caused a large reduction in the free energy of binding (deltadeltaG = +2.8 kcal/mol). Hydrogen bonds provided by (Tyr)L32 and (Asp)H56 were also important for binding: mutation of (Tyr)L32 to Phe resulted in a significant reduction in binding affinity (deltadeltaG = +1.7 kcal/mol). Disruption of hydrophobic interactions (van der Waals contacts) led to significant reductions in affinity also ((Tyr)H99 to Ala, deltadeltaG = +1.5 kcal/mol; (Leu)L94 to Ala, deltadeltaG > +3.0 kcal/mol). An attempt to increase binding affinity by filling a cavity in the interface with a larger antibody side chain was unsuccessful, as the free energy gained by new antibody-antigen interactions did not compensate for the removal of cavity-bound water molecules.

Nonspecific amine immobilization of ligand can Be a potential source of error in BIAcore binding experiments and may reduce binding affinities

The interaction of monovalent forms of NC41, an anti-viral neuraminidase antibody, and the antiidiotype antibody 11-1G10 has been used as a model system for BIAcore analysis to demonstrate the potential problems resulting from the nonspecific amine coupling procedure. To avoid complications due to antibody bivalency, monovalent Fab fragments and monomeric recombinant scFvs were used. When immobilized by amine coupling, the 11-1G10 anti-idiotype fragments were found to have an artificially reduced affinity for NC41 compared to the results obtained using site-directed immobilization via C-terminal thiol residue and from solution equilibrium measurements. The NC41 antibody fragments, on the other hand, were able to retain their 11-1G10 binding affinity when immobilized nonspecifically through free amine groups. These data, in combination with the known sequences of the two antibodies, suggested that nonspecific immobilization through one or more lysine residues close to or within the CDR2 region of the 11-1G10 VH domain was responsible for the reduced strength of the interaction with NC41. These results emphasize the need to use site-specific immobilization strategies when accurate kinetic measurements are required.

Linear gene fusions of antibody fragments with streptavidin can be linked to biotin labelled secondary molecules to form bispecific reagents

Monomeric single chain antibody (scFv) fragments lack both the avidity of the bivalent IgG, or (Fab')2 fragment, and the effector functions conferred by the Fc domain. For certain diagnostic or therapeutic applications it may be desirable to link these molecules to other proteins, antibodies, enzymes or peptide ligands, and chemical or recombinant methods have been developed to produce many of these crosslinked reagents. One approach has been to link an antibody fragment to streptavidin which can bind a second biotinylated molecule to create a higher affinity, bifunctional or bispecific molecule. To demonstrate the applicability of this technology, an anti-neuraminidase NC10 scFv-streptavidin fusion was expressed in E. coli and the product was refolded and purified to homogeneity from 6 M guanidine hydrochloride. Analysis in a BIAcore biosensor showed that the NC10 scFv moiety reacted with immobilised neuraminidase and that the core streptavidin moiety was able to bind biotinylated anti-ferritin Fab' to produce a new model bispecific reagent which bound ferritin. Conceptually, this design principle can be applied to the creation of useful diagnostic and possibly therapeutic molecules.

The 1.8 A crystal structure of winged bean albumin 1, the major albumin from Psophocarpus tetragonolobus (L.) DC

Winged bean albumin-1 (WBA) is the main seed albumin of Psophocarpus tetragonolobus, a legume that has excellent potential as a protein-rich food source for humid tropical climates. WBA crystallises in a tetragonal space group and the structure was solved by X-ray crystallography with a combination of multiple isomorphous replacement using four heavy atom derivatives and molecular replacement with a model based on the structure of Erythrina caffra trypsin inhibitor (ETI). Refinement of the structure proceeded to 1.8 A. WBA has a beta-trefoil fold, similar to that found in the STI-Kunitz type trypsin inhibitors. The final structure has an overall R-factor of 19% for 15 to 1.8 A resolution data, all residues in the allowed regions of the Ramachandran plot, and good agreement with ideal geometry. WBA has sequence similarity with the STI-Kunitz trypsin inhibitors, including the apparent conservation of the functional reactive site residue, lysine 64, at the position of the scissile bond (position P1) in the STI-Kunitz type trypsin inhibitors, however, WBA does not inhibit trypsin. The reason for the lack of inhibitory activity against trypsin is clearly evident from the structure. The loop corresponding to the inhibitory loop in the STI-Kunitz trypsin inhibitors does not conform to the canonical conformation of the inhibitory loops of the "small inhibitors". The lysine residue assigned to the P1 position from sequence alignments is instead part of a four amino acid insertion between residues structurally equivalent to residues P1 and P2 of the inhibitors.

Triabodies: single chain Fv fragments without a linker form trivalent trimers

A single chain Fv fragment (scFv) of the murine monoclonal antibody 11-1G10 was constructed by directly joining the C-terminal residue of the V(H) domain to the N-terminal residue of V(L). 11-1G10 is an anti-idiotype and competes with the antigen, influenza virus neuraminidase (NA), for binding to the NC41 antibody. The scFv formed stable trimers with three active antigen combining sites for NC41 Fab fragments. We propose that trimeric scFvs may be the preferred conformation for directly linked V(H)-V(L) molecules, which contrasts the formation of scFv dimers (diabodies) when the V(H) and V(L) domains are joined by short flexible linkers of between 5-10 residues. BIAcore biosensor binding experiments showed that the trimeric scFv showed an expected increase in binding affinity, due to avidity, compared to the monomeric 15-residue linked scFv. The increase in avidity of scFv trimers offers advantages for imaging and immunotherapy.

Influence of mass transfer and surface ligand heterogeneity on quantitative BIAcore binding data. Analysis of the interaction of NC10 Fab with an anti-idiotype Fab'

The interaction of monovalent Fab fragments of NC10, an antiviral neuraminidase antibody, and the anti-idiotype antibody 3-2G12 has been used as a model system to demonstrate experimentally the influence of non-ideal binding effects on BIAcore binding data. Because the association rate constant for these two molecules was found to be relatively high (about 5 x 10(5) M-1 S-1), mass transfer was recognised as a potential source of error in the analysis of the interaction kinetics. By manipulation of the flow rate and the surface density of the immobilised ligand, however, the magnitude to this error was minimised. In addition, the application of site-specific immobilisation procedures was found to improve considerably the correlation of experimental binding data to the ideal 1:1 kinetic model such that the discrepancy between experimental and fitted curves was within the noise range of the instrument. Experiments performed to measure the equilibrium constant (KD) in solution resulted in a value of similar magnitude to those obtained from the ratio of the kinetic rate constants, even those measured with a heterogeneous ligand or with a significant mass transfer component. For this system, the experimental complexities introduced by covalent immobilisation did not lead to large errors in the KD values obtained using the BIAcore.

Single-chain Fv fragments of anti-neuraminidase antibody NC10 containing five- and ten-residue linkers form dimers and with zero-residue linker a trimer

Single-chain variable fragments (scFvs) of anti-neuraminidase antibody NC10 were constructed by joining the VH and VL domains with 10-residue (Gly4Ser)2 and five-residue (Gly4Ser) linkers; a zero-residue linker scFv was constructed by joining the C-terminal residue of the VH domain to the N-terminus of the VL domain. The scFv with the 10- and five-residue linkers exclusively formed dimeric antibody fragments (M(r) 52000). These were shown to be bivalent and were able to cross-link two neuraminidase tetramers to form a 'sandwich' type complex; each antigen combining site could also bind an anti-idiotype Fab'. The zero-residue linker scFv (M(r) 70000) was shown to form a trimer with three active antigen combining sites, each binding an anti-idiotype Fab' to yield a complex of M(r) 212000. The orientation of the combining sites in the zero-residue linker scFv, however, was such that it could not cross-link tetramers of neuraminidase. BIAcore biosensor experiments showed that the affinity of each individual antigen combining site in both the 10- and five-residue linker scFv dimers and zero-residue linker scFv trimer was essentially the same when the scFvs were immobilized onto the sensor surface. However, when the scFvs were used as the analyte, the dimeric and trimeric scFvs showed an apparent increase in binding affinity due to the avidity of binding the multivalent scFvs.

Identification and minimization of nonideal binding effects in BIAcore analysis: ferritin/anti-ferritin Fab' interaction as a model system

The interaction of human spleen ferritin with a monoclonal antibody Fab' fragment has been studied as a model system for BIAcore analysis. In particular, the influence of nonideal binding effects has been examined both experimentally and by the theoretical simulation of sensorgram curves. Mass transfer effects were found to have a small but significant influence on the observed binding kinetics of the ferritin/antiferritin Fab' interaction; however, this nonideal behavior could be overcome by systematic manipulation of experimental conditions such as the flow rate and the surface density of the immobilized antigen. Because of the multivalent nature of ferritin with 12 antiferritin Fab' binding sites per molecule, immobilization of the antigen by amine coupling had little effect on the majority of free binding sites on the molecule. Consequently, the binding data for both ferritin and apoferritin correlated well with an ideal binding model which assumes binding homogeneity. On the other hand, when ferritin was dissociated to its subunit dimer form (containing one Fab' binding site) prior to surface immobilization significant deviation from this model was observed. This nonideal behavior was probably due to heterogeneity of the immobilized ferritin subunit dimer on the sensor surface, resulting from the nonspecific amine coupling procedure.

Design and expression of a stable bispecific scFv dimer with affinity for both glycophorin and N9 neuraminidase

We have designed and produced a stable bispecific scFv dimer (bisFv) by non-covalent association of two hybrid VH-VL pairs derived from an anti-neuraminidase antibody (NC10) and an anti-glycophorin antibody (1C3). The bisFv dimer was demonstrated to have binding activity to the two respective target antigens and was evaluated as a reagent for rapid whole blood agglutination assays. The bisFv was expressed in the periplasm of Escherichia coli, from a secretion vector which comprised two cistrons in tandem under the control of a single lac promoter, inducible with IPTG. Each cistron encoded one of the hybrid VH-VL pairs, with V domains separated by a linker region encoding the five amino acids, Gly4Ser. The short linker region was designed to prevent association of VH and VL regions of the same molecule and favour the formation of dimers. The protein synthesized from each hybrid scFv cistron was directed to the E. coli periplasm by the inclusion of distinctive signal secretion sequences preceding each hybrid gene; from pel B of Erwinia cartovora and from gene III of fd phage. The bisFv was affinity-purified from culture supernatants via the C-terminal tag epitope FLAG and was shown, by FPLC on a Superose 6 column, to be consistent in size with that of a scFv dimer. The bisFv was stable for more than 4 months at 4 degrees C and was shown by BIAcore analysis to bind to either target antigen, human glycophorin, or tern N9 neuraminidase. Simultaneous binding to both target antigens was demonstrated when a pre-formed bisFv-neuraminidase complex was shown to bind to immobilized glycophorin. In whole blood agglutination assays, the bisFv dimer was able to agglutinate red blood cells when crosslinked with an anti-idiotype antibody (3-2G12) binding to the NC10 combining site, but no agglutination occurred on binding the antigen neuraminidase. These results are a function of the topology of the epitopes on neuraminidase and have implications for the use of relatively rigid bifunctional molecules (as bisFv dimers) to cross link two large membrane-anchored moieties, in this case, red blood cell glycophorin and neuraminidase, an M(r) 190,000 tetramer.

Affinity maturation of recombinant antibodies using E. coli mutator cells

BACKGROUND

Phage libraries can display repertoires of antibodies which are greater in number than the mammalian immune response. However, the selected antibodies often have low binding affinity to their target antigen or hapten (KD below 10(-6) M), which is characteristic of the primary immune repertoire. There is a need for procedures to mimic somatic hypermutation through antigen driven affinity maturation, thereby increasing the affinity of selected immunoglobulins.

OBJECTIVE

To investigate the effectiveness of mutation and affinity selection of recombinant antibody genes with mutator E. coli cells, incorporating phage-display strategies.

STUDY DESIGN

Unique human scFvs were selected from a naive Fd-phage library. These genes were mutated by propagation in mutD5 mutator E. coli cells (mutD5-FIT) which were competent for Fd (M13) based phagemid transfections and generated point mutations (transversions and transitions) in the scFv genes. Individual phage-displayed scFvs were affinity selected from the mutation library and were assayed as soluble scFvs by ELISA and BIAcore for binding to antigen.

RESULTS

The in vivo mutation of phage-displayed scFvs in E. coli mutD5-FIT, combined with affinity selection against antigen, produced scFv molecules with improved binding activity. The point mutations which resulted in single amino acid substitutions frequently produced ten fold increases in apparent binding affinity. Structural comparisons revealed that these point mutations were in framework regions (adjacent to the CDRs) and within the CDRs. In one case the apparent affinity of an anti-glycophorin scFv after mutation in the VL framework region close to CDR3 increased by 10(3). However, this increase in apparent affinity was accompanied by an increased propensity to dimerise and form aggregates.

CONCLUSIONS

A strategy for the rapid affinity maturation of scFv and Fab antibody fragments has been developed which utilises mutator strains of E. coli and incorporates phage display of antibody repertoires (libraries).

A single amino-acid change between the antigenically different extracellular serine proteases V2 and B2 from Dichelobacter nodosus

Dichelobacter nodosus (Dn), the causative organism of ovine footrot, secrets three distinct types of extracellular serine proteases which have been implicated in virulence. Southern analyses have shown that the proteases are encoded by three separate genes, and the genes encoding an acidic protease V5 and a basic protease have already been characterised from virulent Dn strain 198. The gene encoding the third protease type, as represented by acidic protease V2, was isolated from an EcoRI-BamHI library of strain 198 genomic DNA by probing with a polymerase chain reaction (PCR) fragment generated with oligodeoxyribonucleotides based on protease V2 amino acid (aa) sequences. A further clone from an RsaI library was isolated to complete the 5' region of the gene to yield an ORF of 1803 bp encoding a protein precursor of 601 aa. The acidic protease V2 gene, aprV2, shows the same precursor structure as the bprV and aprV5 genes with 72% and 69% similarity at the nucleotide (nt) level and with 73% and 69% similarity at the aa level, respectively. As monoclonal antibodies consistently distinguish the virulent (V) and benign (B) forms of this protease, the gene encoding the acidic protease B2 from benign Dn strain 305 was isolated using the PCR and characterized to investigate the molecular basis for this difference in antigenicity. A 2-bp substitution in a single codon was identified which appeared to be responsible for a change of epitope.

Nucleotide and deduced protein sequence of the extracellular, serine basic protease gene (bprB) from Dichelobacter nodosus strain 305: comparison with the basic protease gene (bprV) from virulent strain 198

In earlier studies, it appeared that benign strains of the Gram-negative, obligate anaerobe, Dichelobacter nodosus, were devoid of the extracellular, serine basic protease (pI approximately 9.5) of virulent strains. However, Southern and PCR analysis have shown a homologous gene (bprB) in the representative benign strain 305. The deduced amino acid sequence of the prepro- and mature protease regions of bprB confirmed this homology and showed 97% sequence identity with the bprV precursor from virulent strain 198. Identity in the carboxy-terminal extension region was 90%. Expression studies in Escherichia coli transformed with bprB, showed that the gene was capable of the production of an active protease. A protease, albeit with a lower iso-electric point (approximately 8.6), was isolated from D. nodosus culture supernatants and shown to cross-react with antibodies raised against the more basic protease from strain 198. The amino acid sequence encoded by the strain 305 gene revealed two additional acidic residues consistent with a lowered iso-electric point and supported the conclusion that bprB and bprV produce equivalent basic proteases.

Solution properties of Escherichia coli-expressed VH domain of anti-neuraminidase antibody NC41

The VH domain of anti-influenza neuraminidase antibody NC41, with and without a C-terminal hydrophilic marker peptide (FLAG), has been expressed in high yield (15-27 mg/L) in Escherichia coli. Both forms were secreted into the periplasm where they formed insoluble aggregates which were solubilized quantitatively with 2 M guanidine hydrochloride and purified to homogeneity by ion-exchange chromatography. The VH-FLAG was composed of three isoforms (pI values of approximately 4.6, 4.9, and 5.3) and the VH molecule was composed of two isoforms with pI values of 5.1 and 6.7; the difference between the VH isoforms was shown to be due to cyclization of the N-terminal glutamine residue in the pI 5.1 isoform. At 20 degrees C and concentrations of 5-10 mg/ml the VH domain dimerized in solution and then partly precipitated, resulting in the broadening of resonances in its 1H NMR spectrum. Reagents such as CHAPS, n-ocytylglucoside, and ethylene glycol, which presumably mask the exposed hydrophobic interface of the VH molecule, prevented dimerization of the VH and permitted good-quality NMR spectra on isotope-labeled protein to be obtained.

Properties of the extracellular acidic proteases of Dichelobacter nodosus. Stability and specificity of peptide bond cleavage

Dichelobacter nodosus, a Gram negative obligate anaerobe and causative organism of ovine footrot, secretes a family of extracellular acidic serine proteases with pI's in the range of 5.2 to 5.6, and a basic serine protease with a pI of approximately 9.5. The acidic proteases show optimum activity at pH 8 and require a divalent metal ion (eg. Ca) to maintain structural integrity. In the presence of EDTA or conditions that cause protein unfolding, the proteases undergo rapid and complete autolysis. The proteases were stable to heating to about 50 degrees C for 30 min but at higher temperatures, activity was rapidly lost; virulent proteases V1 and V2 were slightly more stable (by about 5 degrees C) than benign proteases B2 and B3. The effect of various protease inhibitors on the D. nodosus acidic proteases was the same except that the inhibitor, chymostatin, markedly inhibited protease V5 but not proteases V1-3 or B1-B4. Cleavage of the oxidized insulin B-chain showed that the specificity of proteases V1-V3 and B1-B4 was identical but that it was distinct from that of proteases V5/B5.

Purification of the extracellular acidic proteases of Dichelobacter nodosus

Dichelobacter nodosus, a Gram negative obligate anaerobe and causative organism of ovine footrot, secretes a family of extracellular acidic serine proteases with pI's in the range of 5.2 to 5.6, and a basic serine protease with a pI of approximately 9.5. Four acidic proteases (V1, V2, V3 and V5) from virulent and five acidic proteases (B1 to B5) from benign strains of D. nodosus were purified by chromatography on Sephadex G-100 and DEAE-Sepharose CL-6B. Proteases V2, V5 and B5 were found to yield two forms (a and b) on purification which probably arise from limited autolysis of the parent molecule. Amino acid compositions, peptide profiles produced on autolysis and apparent Mr on SDS-PAGE of proteases V1-V3 showed that they were similar to each other and to proteases B1 to B4, and that these proteases were clearly distinct from proteases V5 and B5, which were found to be identical.

Expression in Escherichia coli of the extracellular basic protease from Dichelobacter nodosus

Dichelobacter nodosus, a Gram-negative obligate anaerobe and the causative agent of ovine footrot, secretes a number of extracellular proteases, one of which is highly basic in nature. The gene (bprV) encoding this basic protease, from virulent strain 198, has been cloned and sequenced. Clone pBR3KB contained the complete bprV gene which constitutively expressed an active protease using its own promoter, when cloned in Escherichia coli. However, levels of protease expression were low and unstable when the clone was expressed in liquid culture. A range of E. coli strains were examined for stable expression; strains NH274 and SURE were found to be better hosts for stable expression than other commonly used E. coli host strains. Stabilization and enhancement of expression was achieved by deletion of the native promoter region and expression from plasmid promoter or promoters, and by modification of culture conditions. The recombinant protease obtained from E. coli was indistinguishable from the native enzyme in size, activity, isoelectric point and immunological properties.

Characterization of a basic serine proteinase (pI approximately 9.5) secreted by virulent strains of Dichelobacter nodosus and identification of a distinct, but closely related, proteinase secreted by benign strains

An extracellular serine proteinase with a PI approximately 9.5 (referred to as 'basic proteinase') was purified to homogeneity, from strains of Dichelobacter nodosus that cause virulent foot-rot, by gel filtration of concentrated culture supernatant on Sephadex G-100 and chromatography on sulphopropyl-Sephadex C-25 at pH 8.6 D. nodosus strains that cause benign foot-rot do not secrete a corresponding basic proteinase with a pI of approximately 9.5. Benign strains secrete a closely related, but distinct, proteinase which has the same molecular mass and N-terminal sequences as the 'virulent' basic proteinase, but a lower pI of approximately 8.6. The basic proteinases from both strains appear to interact with other proteins present in the culture medium, which results in anomalous behavior on gel filtration. Pure D. nodosus 'virulent' basic proteinase has a molecular mass of 36 kDa and showed a low solubility at I < 0.05 precipitating quantitatively from solution as microcrystals. The proteinase shows optimal activity at pH 8.0 and is stable to heating to 55 degrees C for 30 min, but at higher temperatures activity is rapidly lost. Bivalent-metal ions (e.g. Ca2+) are required to maintain the structural integrity and stability of the proteinase; in the presence of EDTA or conditions that cause protein unfolding, the proteinase undergoes rapid and complete autolysis. Cleavage of oxidized insulin A- and B-chain showed that the basic proteinase has a broad specificity, including cleavage at lysine and arginine bonds.

Recombinant anti-sialidase single-chain variable fragment antibody. Characterization, formation of dimer and higher-molecular-mass multimers and the solution of the crystal structure of the single-chain variable fragment/sialidase complex

The single-chain antibody variable fragment (scFv), with a 15-residue polypeptide linker (Gly4Ser)3, of monoclonal antibody NC10 was expressed in Escherichia coli and purified to homogeneity. This scFv molecule, refolded from 6 M guanidine hydrochloride, was predominantly a monomer of 27 kDa and was stable on storage at 4 degrees and 20 degrees C. At higher protein concentrations (approximately 5 mg/ml) dimer and higher-molecular-mass multimers were formed and freezing enhanced this aggregation. The dimer was not stable and dissociated to monomer at 20 degrees C with a half-life of approximately 8 days. The higher-molecular-mass multimers and dimer dissociated to monomer in 60% ethylene glycol. Both the monomer and dimer were active and with tern N9 sialidase yielded complexes of 276 kDa and 569 kDa, respectively, indicating that four scFv molecules bound/sialidase tetramer and that the dimer was bivalent and cross-linked two sialidase tetramers. Binding studies at low concentrations and using radiolabelled scFv indicated that the binding affinity of the dimer was approximately twofold higher than that of the monomer, and the binding affinities of the scFv were similar to that of the parent NC10 antigen-binding fragment (Fab) molecule. A complex between tern N9 sialidase and NC10 scFv was crystallized and the structure of the complex was solved at 0.3-nm resolution by X-ray diffraction. Comparison of this scFv/sialidase structure with the parent Fab/sialidase structure revealed that the modes of attachment of scFv and Fab to sialidase were very similar. There was no discernible electron density for the peptide linker joining the variable heavy (VH) and variable light (VL) chains. A close interaction between two symmetry-related scFv suggests that they may have crystallized as dimers.

Cloning, sequence and expression of the gene (aprV5) encoding extracellular serine acidic protease V5 from Dichelobacter nodosus

The acidic protease V5-encoding gene (aprV5) from Gram- Dichelobacter nodosus virulent strain 198 was isolated from a cosmid bank by activity screening and sequenced. The 2371-bp nucleotide (nt) sequence contained an open reading frame coding for a protein precursor of 595 amino acid (aa) residues composed of a signal peptide, a pro-region, a mature active protease of 347 aa and a C-terminal extension region of 120 aa. The deduced aa sequence of the pre-pro-mature protease regions showed about 65% similarity to that of D. nodosus basic protease while the C-terminal extension region showed only about 26% similarity. The aprV5 gene, without its C-terminal extension region, was cloned and expressed in Escherichia coli. The acidic protease B5-encoding gene (aprB5) from non-virulent strain 305 was also cloned and sequenced. The aprB5 nt sequence showed 99% homology to that of aprV5 with two single-aa changes occurring in the precursor.

Determination of relative binding affinity of influenza virus N9 sialidases with the Fab fragment of monoclonal antibody NC41 using biosensor technology

The relative binding affinities of influenza virus N9 sialidase from term and whale with the Fab fragment of monoclonal antibody NC41 were determined using biosensor technology (Pharmacia BIAcoreTM). The apparent association and dissociation rate constants were measured in real time for the interaction of the Fab with both sialidases, the Fab being immobilised on the sensor surface. Although three-dimensional structural studies have shown that there are no apparent structural differences between the term and whale N9 sialidase epitopes to which the NC41 Fab binds, the apparent binding constant for the interaction with tern N9 sialidase was approximately 2.4-fold higher than that with whale N9 sialidase. The kinetic analysis showed that the association rate constant for the binding of whale N9 sialidase was higher than that for tern N9 sialidase (12.0 x 10(4) M-1 s-1 compared to 4.3 x 10(4) M-1 s-1) and the dissociation rate constants for the whale N9-sialidase-Fab complex were approximately 6-fold higher than for the tern N9-sialidase-Fab complex. Furthermore, kinetic analysis of the dissociation reaction showed that it was composed of two stages, an initial, faster rate followed by a late, slower rate. The values of the relative affinity constants calculated using the initial dissociation rate constant were similar to the values measured at equilibrium in the BIAcore and those determined in true solution equilibrium studies using sedimentation equilibrium. The late, slower, dissociation rate constant yielded affinity constants significantly higher than those obtained by true solution methods.

Use of a gram- signal peptide for protein secretion by gram+ hosts: basic protease of Dichelobacter nodosus is produced and secreted by Bacillus subtilis

The bprV gene, encoding the extracellular basic protease of the Gram- anaerobic bacterium Dichelobacter nodosus, was expressed and the protein secreted in Bacillus subtilis using the novel cloning/expression vector pNC3 [Wu et al., Gene 106 (1991) 103-107]. The pre- and pro-peptides were processed correctly in this heterologous system, and the 127-amino acid C-terminal extension region was also removed. The recombinant gene product was indistinguishable biochemically or immunochemically from the authentic protease and was able to form crystals upon dialysis, as was found for the authentic protease. This is the first example of the direct secretion of a Gram- extracellular enzyme in B. subtilis via its own signal peptide. The fact that this gene can be expressed and its product secreted in both Escherichia coli and B. subtilis provides a unique opportunity to study and compare the similarities and differences in protein secretion between Gram- and Gram+ organisms.

Characterisation of the wheat Mr 15000 "grain-softness protein" and analysis of the relationship between its accumulation in the whole seed and grain softness

The Mr 15000 protein associated with water-washed wheat starch granules from soft wheats was shown to be heterogeneous: it could be divided into a fraction containing one or moreα-amylase inhibitor subunits and a fraction largely composed of a previously uncharacterised polypeptide(s) referred to as the "grainsoftness protein" (GSP). The major N-terminal sequence and sequences of peptides derived from protease digests of GSP are reported. An antiserum specific for GSP was used to show that GSP accumulated in both hard and soft wheat grains, but the GSP in soft grains associated more strongly with starch granules than the GSP in hard grains. A positive correlation between grain softness and accumulation of GSP in the seed was demonstrated for a range of cultivars. This differs from the qualitative relationship, based on the isolated starch fraction, between GSP and grain softness that has already been reported. Analysis of wholemeal extracts with the antiserum demonstrated that the accumulation of GSP in the seed was dependent on the short arm of chromosome 5D, which also encodes theHa locus. In addition, examination of near-isogenic lines differing in hardness indicated that the gene(s) controlling GSP was (were) linked with theHa locus. The findings indicate that GSP may be the product of theHa locus and thus be the major factor that determines the milling characteristics of bread wheats.

Recombinant antineuraminidase single chain antibody: expression, characterization, and crystallization in complex with antigen

The variable heavy (VH) and variable light (VL) genes of NC10, a monoclonal antibody with specificity toward N9 neuraminidase (NA), were cloned and sequenced. A single chain Fv (scFv) fragment of NC10, consisting of VH and VL domains joined by a peptide linker, was designed, constructed and expressed in the E. coli expression vector pPOW. The N-terminal secretion signal PelB directed the synthesized protein into the periplasm where it was associated with the insoluble membrane fraction. An octapeptide (FLAG) tail was fused to the C-terminus of the single chain Fv to aid in its detection and remained intact throughout the protein purification process. NC10 scFv was purified by solubilization of the E. coli membrane fraction with guanidinium hydrochloride followed by column chromatography. The purified NC10 scFv showed binding affinity for its antigen, NA, 2-fold lower than that of the parent Fab. The complex between NA and the scFv has been crystallized by the vapor diffusion method. The crystals are tetragonal, space group P42(1)2, with unit cell dimensions a = b = 141 A, c = 218 A.

Amino acid sequence of extracellular acidic protease V5 of Dichelobacter nodosus, the causative organism of ovine footrot

Dichelobacter nodosus, a Gram-negative obligate anaerobe and the causative organism of ovine footrot, secretes a family of extracellular serine proteases with pI's in the range of 5.2 to 5.6 and a serine basic protease with a pI of approximately 9.5. The primary structure of acidic protease V5 (pI approximately 5.2) from D. nodosus virulent strain 198 was determined by direct amino acid sequencing. This protease consists of a single polypeptide chain of 347 amino acids, contains two disulfide bonds and has a M(r) of 35960. Comparison of the D. nodosus acidic protease V5 sequence with that of other serine proteases showed that it is a member of the subtilisin family of proteases with strong conservation of identity around the catalytic residues. The sequence of protease V5 showed 64% identity to D. nodosus basic protease (pI approximately 9.5) and 53% identity to the extracellular serine protease of Xanthomonas campestris, a plant pathogen but only 25-35% identity to other proteases of the subtilisin family. The D. nodosus proteases are similar in length to X. campestris protease (but some 70 residues shorter than the subtilisins) and they share two conserved disulfide bonds with the X. campestris protease, a feature not observed for other members of the subtilisin family.

Amino acid and DNA sequences of an extracellular basic protease of Dichelobacter nodosus show that it is a member of the subtilisin family of proteases

A DNA fragment encoding an extracellular basic protease (pI approximately 9.5) from Dichelobacter nodosus, a Gram-negative obligate anaerobe and the causative agent of ovine footrot, has been cloned and expressed in Escherichia coli and sequenced. E. coli harbouring a plasmid with a 3-kb DNA fragment containing the D. nodosus basic-protease gene exhibited proteolytic activity when tested on skim-milk plates. The sequence of the native basic protease isolated from D. nodosus was also determined by direct amino acid sequencing. Comparison of the deduced sequence of the primary translation product (603 residues) and that of the native protease (344 residues) indicates that the protease is synthesized as a precursor molecule, containing a signal peptide (21 residues), a 111 amino acid pro-peptide and a 127 residue C-terminal extension which is subsequently processed to the mature active form. Comparison of the D. nodosus basic protease sequence with that of other serine proteases showed that it is related to the subtilisin family of proteases with strong conservation of sequence identity around the catalytic site residues. A remarkable similarity in structure was found to the serine protease of Xanthomonas campestris, a plant pathogen, with respect to the length of the precursor segments, conservation of disulfide bridges and approximately 50% sequence identity of the mature proteases.

The structure of the complex between influenza virus neuraminidase and sialic acid, the viral receptor

Crystallographic studies of neuraminidase-sialic acid complexes indicate that sialic acid is distorted on binding the enzyme. Three arginine residues on the enzyme interact with the carboxylate group of the sugar which is observed to be equatorial to the saccharide ring as a consequence of its distorted geometry. The glycosidic oxygen is positioned within hydrogen-bonding distance of Asp-151, implicating this residue in catalysis.

High-level temperature-induced synthesis of an antibody VH-domain in Escherichia coli using the PelB secretion signal

We have constructed a temperature-inducible Escherichia coli expression vector (pPOW) for enhanced secretion of antibody (Ab) domains and other foreign proteins. The vector contains the lambda pRpL promoters in tandem, and the cI857 gene encoding the temperature-sensitive repressor which provide tight control over protein production. The PelB secretion signal directs the synthesized foreign protein through the cytoplasmic membrane. A mouse Ab fragment (the variable heavy (VH) domain of NC41) was synthesized efficiently by this vector and accumulated with the cell membranes (not as inclusion bodies) at levels of 30 mg/l. This represents the highest yields reported to date for Ab fragments with a native N terminus. An octapeptide (FLAG) tail was fused to the C terminus of the VH domain to aid in purification, and remained intact throughout the protein purification process. The optimum conditions for protein production were controlled by the type of culture medium used, the age of the bacterial population at the time of induction, and the period of synthesis of the protein product. The purified Ab VH fragment showed binding affinity (Ka less than 10(4)/M) to its target antigen (neuraminidase).

Primary structure of Kunitz-type trypsin inhibitor-2a (pI 5.9) from Psophocarpus tetragonolobus (L.) DC seed

The primary structure of acidic trypsin inhibitor-2a (WBTI-2a, pI 5.9) from Psophocarpus tetragonolobus (L.) DC seed was determined. This inhibitor consists of a single polypeptide chain of 180 amino acids including four half-cystine residues and has an N-terminal residue of pyroglutamic acid. The sequence of WBTI-2a, pI 5.9, showed 84% identity to acidic trypsin inhibitor-2 (WBTI-2, pI 5.1) but only 57% identity to the basic trypsin inhibitor (WBTI-1, pI 8.9) and 50% identity to the chymotrypsin inhibitor of winged bean. The data indicate that winged bean seed contains a family of three Kunitz-type inhibitors which have about 50% identity.

A new method for the purification of the influenza A virus neuraminidase

A rapid new method for the purification of neuraminidase (NA) heads from influenza A virus is described. Virus was pelleted directly from allantoic fluid and was digested with pronase. The cores were removed by centrifugation, redigested and the released NA heads were pooled and concentrated. The NA was separated from all contaminating proteins in a single step on a Superose 12 column. The purified material was suitable for both crystallography and for the production of monospecific antisera.

Amino acid and cDNA sequences of a methionine-rich 2S protein from sunflower seed (Helianthus annuus L.)

The amino acid sequence of a methionine-rich 2S seed protein from sunflower (Helianthus annuus L.) and the sequence of a cDNA clone which codes for the entire primary translation product have been determined. The mature protein consists of a single polypeptide chain of 103 amino acids (molecular mass 12133 Da) which contains 16 residues of methionine and 8 residues of cysteine. The cDNA sequence established that the protein is synthesized as a precursor of 141 residues with a typical hydrophobic signal sequence of 25 residues followed by a further 13-residue hydrophobic pro-sequence which is presumably removed by post-translational cleavage. The sequence of the mature protein and that deduced from the cDNA were identical with no evidence of processing at the C-terminus. Comparison of the sunflower methionine-rich protein sequence with sequences of other seed 2S proteins from dicotyledons and monocotyledons showed limited but distinct sequence similarities; in particular the arrangement of the cysteine residues was conserved. The sunflower protein shows 34% identity with the methionine-rich Brazil nut 2S protein and the prepro regions of the precursors of these two proteins show about 50% identity. This similarity indicates that these methionine-rich 2S proteins have diverged as a subclass of the 2S superfamily of proteins which contain only 2-3% methionine. While the related 2S proteins from other dicotyledons are processed to a small and large subunit, the sunflower protein is not cleaved in this way.

Amino acid sequence and disulfide bridges of affinity purified Kunitz-type chymotrypsin inhibitor from winged bean seed (Psophocarpus tetragonolobus (L.) DC)

The primary sequence of the affinity purified chymotrypsin inhibitor, WBCI, isolated from the albumin fraction of Psophocarpus tetragonolobus (L.) DC cv. UPS-122 seed was determined. The inhibitor consisted of a single polypeptide chain of 183 amino acids (Mr 20285) and the four half-cystine residues in the molecule formed two intramolecular disulfide bridges equivalent to those in other Kunitz-type seed inhibitors. The sequence of this chymotrypsin inhibitor was identical to that of chymotrypsin inhibitor-3 from cultivar UPS-31 and it showed about 50% sequence similarity to the winged bean acidic (WBTI-2, pI 5.1) and basic (WBTI-1, pI 8.9) trypsin inhibitors. Sequence similarities to other Kunitz-type seed inhibitors are discussed.

Pea convicilin: structure and primary sequence of the protein and expression of a gene in the seeds of transgenic tobacco

Convicilin, a trimeric globulin of pea (Pisum sativum L.) seeds, is closely related to vicilin and composed of polypeptides of 68.2 kilodaltons. A partial copy DNA (cDNA) clone encoding convicilin was isolated, sequenced, and used to select a convicilin gene from a pea genomic library. A part of the genomic clone was sequenced to obtain the coding sequences missing in the cDNA clone and a further 1 kilobase 5' to the start of transcription were also obtained. The entire sequence of convicilin was deduced from the combined genomic and cDNA sequences. The complete gene encoding convicilin was transferred to tobacco (Nicotiana tabacum L.) and the characteristics of its expression in the seeds of transgenic plants were studied. An unprocessed polypeptide, which was found only in the seeds of the transgenic plants, was identical in size to pea convicilin, and was recognized by vicilin antibodies. Convicilin, which does not undergo posttranslational cleavage in peas, was partially processed to polypeptides of a relative molecular mass (Mr) of approx. 50000 in transgenic tobacco seeds. There was a twofold variation in the level of convicilin accumulated by the mature seeds of a number of transgenic plants and this was well correlated with the number of gene copies incorporated in the different transformants. In the seeds of tobacco plants that contained a single copy of the transferred gene it was estimated that convicilin comprised up to 2% of the seed protein. Thus, using a combination of gene sequencing and expression in a heterologous host we believe we have characterized the gene corresponding to theCvc locus, whereas the gene described by D. Bown et al. (1988, Biochem J.,251, 717-726) probably encodes a minor convicilin-related protein.

A cDNA-based comparison of dehydration-induced proteins (dehydrins) in barley and corn

Several cDNAs related to an ABA-induced cDNA from barley aleurone were isolated from barley and corn seedlings that were undergoing dehydration. Four different barley polypeptides with sizes of 22.6, 16.2, 14.4 and 14.2 kDa and a single corn polypeptide with a size of 17.0 kDa were predicted from the nucleotide sequences of the cDNAs. These dehydration-induced proteins (dehydrins) are very similar to each other and to a previously identified rice protein induced by ABA and salt, and have at least some similarity to a previously identified cotton embryo protein. Each dehydrin is extremely hydrophilic, glycine-rich, cysteine- and tryptophan-free and contains repeated units in a conserved linear order. A lysine-rich repeating unit occurs twice in each protein, once at the carboxy terminus and once partway through the polypeptide, adjacent to a succession of serines. This repeating unit and the adjacent flanking run of serines are conserved with minimal variation among all dehydrins. Another repeating unit is flanked by the two copies of the lysine-rich unit, and varies in number from one to five copies. This latter repeating unit is less conserved than the former, varying even within a singly dehydrin. The messenger RNAs corresponding to each cDNA are abundant in dehydrating, but not in well-watered seedlings. The amino acid sequence of tryptic peptides from purified dehydration-induced proteins of corn established that the corn cDNAs correspond to a protein that is produced in abundance during the response of corn seedlings to dehydration.

Amino acid sequence of a crystalline seed albumin (winged bean albumin-1) from Psophocarpus tetragonolobus (L.) DC. Sequence similarity with Kunitz-type seed inhibitors and 7S storage globulins

The complete amino acid sequence of winged bean albumin-1 (WBA-1) of Psophocarpus tetragonolobus (L.) DC has been determined. The protein consists of a single polypeptide chain of 175 amino acid residues, with one disulfide bond, corresponding to a molecular mass of 19333 Da. WBA-1 was found to be homologous with the Kunitz-type seed trypsin inhibitors. The similarity between WBA-1 and the trypsin inhibitors from soybean and winged bean was 38% and 28%, respectively; similarity was most marked in the C-terminal third of the sequence with identities of 47% and 37%, respectively. Significant similarity was found also between the 2S Kunitz-type proteins and the carboxy-terminal region of the 7S storage globulins, suggesting that these two groups of proteins are related and may have evolved from a common ancestral precursor. Circular dichroism measurements suggest a high content of beta sheet (52%) while secondary structure predictions based on amino acid sequence indicate a similar content and distribution of beta sheet to that found for soybean trypsin inhibitor by X-ray diffraction studies.

Amino acid sequences of hemoglobins I and II from root nodules of the non-leguminous Parasponia rigida-rhizobium symbiosis, and a correction of the sequence of hemoglobin I from Parasponia andersonii

The amino acid sequence of hemoglobins I (pI 6.15 as oxyhemoglobin) and II (pI 5.64 as oxyhemoglobin) from the nitrogen-fixing root nodules of Parasponia rigida have been determined by protein sequencing. The sequence of hemoglobin I (pI 6.16, as oxyhemoglobin) from Parasponia andersonii was re-examined and the corrected primary structure, now in agreement with that predicted from the DNA sequence, is reported. The three Parasponia hemoglobins contain 161 amino acid residues (Mr approximately equal to 18,700 including the heme) with a single cysteine residue and five methionine residues. The N-terminal serine is blocked by an acetyl group. The primary structure of the Parasponia hemoglobins is highly conserved. Hemoglobins I from the two species of Parasponia are identical; both show microheterogeneity at position 30 (Asp/Glu substitution) and hemoglobin I from P. rigida shows microheterogeneity at position 150 (Ala/Val) while hemoglobin I from P. andersonii has only an Ala at 150. P. rigida hemoglobin II shows no microheterogeneity at these positions, having Asp and Val residues respectively, and it contains a single amino acid change of a Gln for an Arg at position 85, which accounts for the 0.5 unit difference in isoelectric point observed between hemoglobins I and II. The sequence data are consistent with allelic heterogeneity at a single locus rather than different genes.

Sequence and identification of the barley yellow dwarf virus coat protein gene

The nucleotide sequence of the coat protein gene of barley yellow dwarf virus (BYDV, PAV serotype) was determined, and the amino acid sequence was deduced. The open reading frame, encoding a protein of relative molecular mass (Mr) 22,047, was confirmed as the coat protein gene by comparison with amino acid sequences of tryptic peptides derived from dissociated virions. In addition, a fragment of this gene expressed in Escherichia coli produced a product which was recognized by antibodies prepared against purified BYDV virions. An overlapping reading frame encoding an Mr 17,147 protein is contained completely within the coat protein gene.

Crystallization and preliminary crystallographic data of the major albumin from Psophocarpus tetragonolobus (L.) DC

The major seed albumin from Psophocarpus tetragonolobus (L.) DC, winged bean albumin 1, has been crystallized from ethanol as rhombic needles in a form suitable for high resolution x-ray crystallographic studies. The space group is I4(1)22 with cell parameters a = b = 95.6 A, and c = 86.0 A. Amino acid sequence homology between winged bean albumin 1 and legume 7 S globulins suggests that the former may be related to a structural domain of the 7 S globulins.