Human vascular endothelial growth factor B: characterization of recombinant isoforms and generation of neutralizing monoclonal antibodies.

1. The vascular endothelial growth factor (VEGF) family is a focus of interest with respect to novel therapies for cardiovascular disease. Members of this family bind differentially to three receptor tyrosine kinases, namely VEGF-R1, -R2 and -R3, and to the semaphorin receptors neuropilin 1 and 2. The role of VEGF-R1 and the factors that interact exclusively with this receptor (VEGF-B and placenta growth factor) has remained controversial. 2. To further elucidate the role of VEGF-B in blood vessel formation and function, we have expressed, purified and refolded both naturally occurring VEGF-B isoforms and a truncated amino acid 10-108 form. All refolded proteins have been demonstrated to bind to VEGF-R1 with appropriate kinetics in biosensor-based analysis. 3. Robust cell assays for VEGF-R1 ligands, such as VEGF-B, have been problematic. We have developed an assay based on a chimeric receptor consisting of extracellular domains 1-4 of VEGF-R1 and the transmembrane and intracellular domains of gp130. The cell line expresses luciferase to high levels 24 h after exposure to VEGF-A and both refolded VEGF-B167 and the short 10-108 isoform have been demonstrated to be active in this assay. 4. The novel cell-based assay, in combination with a variety of immunochemical approaches, has been used to identify and characterize monoclonal antibodies that neutralize VEGF-B activity.

Purification and refolding of vascular endothelial growth factor-B.

Vascular endothelial growth factor (VEGF)-A interacts with the receptor tyrosine kinases VEGF-R1 and R2, and the importance of this interaction in endothelial cell (EC) function and blood vessel development has been well documented. Other ligands that interact differentially with these receptors and that are structurally related to VEGF-A include VEGF-B, VEGF-C, VEGF-D, and placenta growth factor (PLGF). Compared with VEGF-A, relatively little is known about the biological role of the VEGF-R1 specific ligand, VEGF-B. Two splice variant isoforms that differ at the COOH-terminus and which retain unique solubility characteristics are widely expressed throughout embryonic and postnatal development. Recent analysis of mice with a targeted deletion of the VEGF-B gene has revealed a defect in heart development and function consistent with an important role in vascularization of the myocardium (Bellomo D et al., 2000, Circ Res 86:E29-E35). To facilitate further characterization of VEGF-B, we have developed a protocol for expression and purification of refolded recombinant protein from Escherichia coli inclusion bodies (IBs). The approach developed resolves a number of significant issues associated with VEGF-B, including the ability to heterodimerize with endogenous VEGF-A when co-expressed in mammalian cells, a complex secondary structure incorporating inter- and intrachain disulfide bonds and hydrophobic characteristics that preclude the use of standard chromatographic resins. The resulting purified disulfide-linked homodimer was demonstrated to bind to VEGF-R1 and to compete with VEGF-A for binding to this receptor.

Dynamics of the metallo-beta-lactamase from Bacteroides fragilis in the presence and absence of a tight-binding inhibitor.

A significant determinant for the broad substrate specificity of the metallo-beta-lactamases from Bacteroides fragilis and other similar organisms is the presence of a plastic substrate binding site that is nevertheless capable of tight substrate binding in the Michaelis complex. To achieve these two competing ends, the molecule apparently employs a flexible flap that closes over the active site in the presence of substrate. These characteristics imply that dynamic changes are an important component of the mechanism of action of these enzymes. The backbone and tryptophan side chain dynamics of the metallo-beta-lactamase from B. fragilis have been examined using (15)N NMR relaxation measurements. Two states of the protein were examined, in the presence and absence of a tight-binding inhibitor. Relaxation measurements were analyzed by the model-free method. Overall, the metallo-beta-lactamase molecule is rigid and shows little flexibility except in loops. The flexibility of the loop that covers the active site is not unusually great as compared to the other loops of the protein. Local motion on a picosecond time scale was found to be very similar throughout the protein in the presence and absence of the inhibitor, but a significant difference was observed in the motions on a nanosecond time scale (tau(e)). Large-amplitude motions with a time constant of about 1.3 ns were observed for the flexible flap region (residues 45-55) in the absence of the inhibitor. These motions were completely damped out in the presence of the inhibitor. In addition, the motion of a tryptophan side chain at the tip of the beta-hairpin of the flap shows a very significant difference in motion on the ps time scale. These results indicate that the motions of the polypeptide chain in the flap region can be invoked to explain both the wide substrate specificity (the free form has considerable amplitude of motion in this region) and the catalytic efficiency of the metallo-beta-lactamase (the motions are damped out when the inhibitor and by implication a substrate binds in the active site).

NMR characterization of the metallo-beta-lactamase from Bacteroides fragilis and its interaction with a tight-binding inhibitor: role of an active-site loop.

Understanding the structure and dynamics of the enzymes that mediate antibiotic resistance of pathogenic bacteria will allow us to take steps to combat this increasingly serious public health hazard. Complete backbone NMR resonance assignments have been made for the broad-specificity metallo-beta-lactamase CcrA from Bacteroides fragilis in the presence and absence of a tight-binding inhibitor. Chemical shift indices show that the secondary structure of the CcrA molecule in solution is very similar to that in published crystal structures. A loop adjacent to the two-zinc catalytic site exhibits significant structural variation in the published structures, but appears from the NMR experiments to be a regular beta-hairpin. Backbone heteronuclear NOE measurements indicate that this region has slightly greater flexibility on a picosecond to nanosecond time scale than the molecule as a whole. The loop appears to have an important role in the binding of substrates and inhibitors. Binding of the inhibitor 3-[2'-(S)-benzyl-3'-mercaptopropanoyl]-4-(S)-carboxy-5, 5-dimethylthiazolidine causes a marked increase in the stability of the protein toward unfolding and aggregation, and causes changes in the NMR resonance frequencies of residues close to the active (zinc-binding) site, including the beta-hairpin loop. There is a small but significant increase in the heteronuclear NOE for this loop upon inhibitor binding, indicative of a decrease in flexibility. In particular, the NOE of the indole ring of tryptophan 49, at the tip of the beta-hairpin loop, changes from a low value characteristic of a random coil chain to a significantly higher value, close to that observed for the backbone amides in this region of the protein. These results strongly suggest that the hairpin loop participates in the binding of substrate and in the shielding of the zinc sites from solvent. The broad specificity of the CcrA metallo-beta-lactamase may in fact reside in the plasticity of this part of the protein, which allows it to accommodate and bind tightly to substrates of a variety of shapes and sizes.

The identification of metal-binding ligand residues in metalloproteins using nuclear magnetic resonance spectroscopy.

The identification of metal-binding ligands in metalloproteins is an important step in gaining detailed information regarding the environment of the active site. Traditionally, techniques such as 13Cd-substitution for the active metal followed by isotope-filtered NMR techniques have been used to this end. However, for medium to high molecular weight proteins (>20 kDa), these experiments may not be beneficial due to extensive 1H spectral overlap. Here, we describe an alternative approach, where metal-binding ligands such as histidine and cysteine are specifically 15N backbone labeled, excess EDTA is added and changes to (1H-15N) HSQC spectra are followed. Under these conditions, the amide groups of all 15N labeled histidine and cysteine residues, which were either ligands or residues close to the active site, were identified unambiguously for metallo-beta-lactamase from Bacteroides fragilis.

Comparison of native and mutant proteins provides a sequence-specific assignment of the cysteinyl ligand proton NMR resonances in the 2[Fe4S4] ferredoxin from Clostridium pasteurianum.

A sequence-specific assignment is presented for the eight low-field paramagnetically shifted cysteinyl ligand proton NMR resonances in the 2[Fe4S4] ferredoxin from Clostridium pasteurianum. The assignment is based upon comparison of chemical shifts in 1D and 2D NMR spectra of native oxidized protein and those of three mutants. The mutant proteins G12A and G41A were designed to produce minor local structural changes (hence small chemical shift perturbations) in either cluster I (glycine 12 to alanine) or in cluster II (glycine 41 to alanine). Observed chemical shift changes in spectra of the double mutant G12,41A support the interpretation. The comparison is aided by structural models derived from the crystal structure of the related ferredoxin from Peptococcus aerogenes. Each of the eight low-field resonances is assigned to a beta-proton from a different cysteinyl ligand, and so connectivities established from previous TOCSY and HMQC data allow assignment of all 24 cysteinyl ligand protons.