Computational design of a specific heavy chain/κ light chain interface for expressing fully IgG bispecific antibodies.

The use of bispecific antibodies (BsAbs) to treat human diseases is on the rise. Increasingly complex and powerful therapeutic mechanisms made possible by BsAbs are spurring innovation of novel BsAb formats and methods for their production. The long-lived in vivo pharmacokinetics, optimal biophysical properties and potential effector functions of natural IgG monoclonal (and monospecific) antibodies has resulted in a push to generate fully IgG BsAb formats with the same quaternary structure as monoclonal IgGs. The production of fully IgG BsAbs is challenging because of the highly heterogeneous pairing of heavy chains (HCs) and light chains (LCs) when produced in mammalian cells with two IgG HCs and two LCs. A solution to the HC heterodimerization aspect of IgG BsAb production was first discovered two decades ago; however, addressing the LC mispairing issue has remained intractable until recently. Here, we use computational and rational engineering to develop novel designs to the HC/LC pairing issue, and particularly for κ LCs. Crystal structures of these designs highlight the interactions that provide HC/LC specificity. We produce and characterize multiple fully IgG BsAbs using these novel designs. We demonstrate the importance of specificity engineering in both the variable and constant domains to achieve robust HC/LC specificity within all the BsAbs. These solutions facilitate the production of fully IgG BsAbs for clinical use.

Structural analysis of a set of proteins resulting from a bacterial genomics project.

The targets of the Structural GenomiX (SGX) bacterial genomics project were proteins conserved in multiple prokaryotic organisms with no obvious sequence homolog in the Protein Data Bank of known structures. The outcome of this work was 80 structures, covering 60 unique sequences and 49 different genes. Experimental phase determination from proteins incorporating Se-Met was carried out for 45 structures with most of the remainder solved by molecular replacement using members of the experimentally phased set as search models. An automated tool was developed to deposit these structures in the Protein Data Bank, along with the associated X-ray diffraction data (including refined experimental phases) and experimentally confirmed sequences. BLAST comparisons of the SGX structures with structures that had appeared in the Protein Data Bank over the intervening 3.5 years since the SGX target list had been compiled identified homologs for 49 of the 60 unique sequences represented by the SGX structures. This result indicates that, for bacterial structures that are relatively easy to express, purify, and crystallize, the structural coverage of gene space is proceeding rapidly. More distant sequence-structure relationships between the SGX and PDB structures were investigated using PDB-BLAST and Combinatorial Extension (CE). Only one structure, SufD, has a truly unique topology compared to all folds in the PDB.

Deposit3D: a tool for automating structure depositions to the Protein Data Bank.

Almost all successful protein structure-determination projects in the public sector culminate in a structure deposition to the Protein Data Bank (PDB). In order to expedite the deposition process, Deposit3D has been developed. This command-line script calculates or gathers all the required structure-deposition information and outputs this data into a mmCIF file for subsequent upload through the RCSB PDB ADIT interface. Deposit3D might be particularly useful for structural genomics pipeline projects because it allows workers involved with various stages of a structure-determination project to pool their different categories of annotation information before starting a deposition session.

Crystallization and preliminary crystallographic data of the PAS domain of the NifL protein from Azotobacter vinelandii.

The Azotobacter vinelandii NifL protein is a redox-sensing flavoprotein which inhibits the activity of the nitrogen-specific transcriptional activator NifA. The N-terminal PAS domain has been overexpressed in Escherichia coli and crystallized by the hanging-drop vapour-diffusion method. The crystal belongs to the rhombohedral space group R32, with unit-cell parameters a = b = 65.0, c = 157.3 A, and has one molecule in the asymmetric unit. Native data were collected to 3.0 A on the BW7B synchrotron beamline at the EMBL Hamburg Outstation.

A novel quaternary structure of the dimeric alpha-crystallin domain with chaperone-like activity.

alphaB-crystallin, a member of the small heat-shock protein family and a major eye lens protein, is a high molecular mass assembly and can act as a molecular chaperone. We report a synchrotron radiation x-ray solution scattering study of a truncation mutant from the human alphaB-crystallin (alphaB57-157), a dimeric protein that comprises the alpha-crystallin domain of the alphaB-crystallin and retains a significant chaperone-like activity. According to the sequence analysis (more than 23% identity), the monomeric fold of the alpha-crystallin domain should be close to that of the small heat-shock protein from Methanococcus jannaschii (MjHSP16.5). The theoretical scattering pattern computed from the crystallographic model of the dimeric MjHSP16.5 deviates significantly from the experimental scattering by the alpha-crystallin domain, pointing to different quaternary structures of the two proteins. A rigid body modeling against the solution scattering data yields a model of the alpha-crystallin domain revealing a new dimerization interface. The latter consists of a strand-turn-strand motif contributed by each of the monomers, which form a four-stranded, antiparallel, intersubunit composite beta-sheet. This model agrees with the recent spin labeling results and suggests that the alphaB-crystallin is composed by flexible building units with an extended surface area. This flexibility may be important for biological activity and for the formation of alphaB-crystallin complexes of variable sizes and compositions.

1.8 and 1.9 A resolution structures of the Penicillium amagasakiense and Aspergillus niger glucose oxidases as a basis for modelling substrate complexes.

Glucose oxidase is a flavin-dependent enzyme which catalyses the oxidation of beta-D-glucose by molecular oxygen to delta-gluconolactone and hydrogen peroxide. The structure of the enzyme from Aspergillus niger, previously refined at 2.3 A resolution, has been refined at 1.9 A resolution to an R value of 19.0%, and the structure of the enzyme from Penicillium amagasakiense, which has 65% sequence identity, has been determined by molecular replacement and refined at 1.8 A resolution to an R value of 16.4%. The structures of the partially deglycosylated enzymes have an r.m.s. deviation of 0.7 A for main-chain atoms and show four N-glycosylation sites, with an extended carbohydrate moiety at Asn89. Substrate complexes of the enzyme from A. niger were modelled by force-field methods. The resulting model is consistent with results from site-directed mutagenesis experiments and shows the beta-D-glucose molecule in the active site of glucose oxidase, stabilized by 12 hydrogen bonds and by hydrophobic contacts to three neighbouring aromatic residues and to flavin adenine dinucleotide. Other hexoses, such as alpha-D-glucose, mannose and galactose, which are poor substrates for the enzyme, and 2-deoxy-D-glucose, form either fewer bonds or unfavourable contacts with neighbouring amino acids. Simulation of the complex between the reduced enzyme and the product, delta-gluconolactone, has provided an explanation for the lack of product inhibition by the lactone.

Mutation and modeling analysis of the Saccharomyces cerevisiae Swi6 ankyrin repeats.

The Swi4-Swi6 family of transcription factors confers G1/S specific transcription in budding and fission yeast. These proteins contain four ankyrin repeats, which are present in a large number of functionally diverse proteins and have been shown to be important for protein-protein interaction. However, no specific sequence has been identified that is diagnostic of an ankyrin repeat-interacting protein. To determine the function of the ankyrin repeats of Swi6, we generated both random and site-directed mutations within the ankyrin repeat domain of Swi6 and assayed the transcriptional function of these mutant swi6 alleles. We found six single mutations, scattered within the first and the fourth repeats, that generate a temperature-sensitive Swi6 protein. In addition, we found that alanine substitutions for the most conserved residues in each repeat were highly deleterious and also confer temperature sensitivity. Most of these swi6 alleles are able to form ternary complexes with Swi4 and DNA, but these complexes display reduced mobility in band-shift gels, suggesting a dramatic conformational change. We have modeled the ankyrin repeats of Swi6 using the coordinates derived for 53BP2 and find that, despite its low level of sequence conservation, these modeling studies and our mutation data are consistent with Swi6 having a structure very similar to that of 53BP2. Moreover, all but one of our single mutants and all of the site-directed mutants disrupt critical structural features of the predicted folding pattern of these repeats. We conclude that the ankyrin repeats play a major structural role in Swi6. Ankyrin repeats are unlikely to have inherent protein or DNA binding properties. However, they form a characteristic and stable structure with surfaces that may be tailored for many different macromolecular interactions.

Structural and biochemical properties of glycosylated and deglycosylated glucose oxidase from Penicillium amagasakiense.

Glucose oxidase from Penicillium amagasakiense was purified to homogeneity by ion-exchange chromatography and deglycosylated with endoglycosidase H. On the basis of gas chromatography and sodium dodecyl sulphate/polyacrylamide gel electrophoretic (SDS-PAGE) analyses, the protein-bound high-mannose-type carbohydrate moiety corresponded to 13% of the molecular mass of glycosylated glucose oxidase. A total of six N-glycosylation sites per dimer were determined from the N-acetylglucosamine content. The enzymatically deglycosylated enzyme contained less than 5% of the original carbohydrate moiety. A molecular mass of 130 kDa (gel filtration) and 133 kDa (native PAGE) was determined for the dimer and 67 kDa (SDS-PAGE) for the monomer of the deglycosylated enzyme. The N-terminal sequence, which has not-been published for glucose oxidase from P. amagasakiense to date and which showed less than 50% homology to the N terminus of glucose oxidase from Aspergillus niger, and the amino acid composition were not altered by the deglycosylation. Deglycosylation also did not affect the kinetics of glucose oxidation or the pH and temperature optima. It also did not increase the susceptibility of the enzyme to proteolytic degradation. However, deglycosylated glucose oxidase exhibited decreased pH and thermal stability. The thermal stability of both enzymes was shown to be dependent on the buffer concentration and was enhanced by certain additives, particularly 1 M (NH4)2SO4, which stabilised glucose oxidase 100- to 300-fold at 50 degrees C and pH 7-8, and 2 M KF, which stabilised the enzyme up to 36-fold at 60 degrees C and pH 6. In sodium acetate buffer, changes in pH (4-6) affected the affinity for glucose but had no effect on the Vmax of the reaction. In contrast, in TRIS buffer, pH 8, a 10-fold decrease in Vmax and a 2-fold decrease in K(m) were observed.

Modified substrate specificity of L-hydroxyisocaproate dehydrogenase derived from structure-based protein engineering.

L-2-Hydroxyisocaproate dehydrogenase (L-HicDH) is characterized by a broad substrate specificity and utilizes a wide range of 2-oxo acids branched at the C4 atom. Modifications have been made to the sequence of the NAD(H)-dependent L-HicDH from Lactobacillus confusus in order to define and alter the region of substrate specificity towards various 2-oxocarbonic acids. All variations were based on a 3D-structure model of the enzyme using the X-ray coordinates of the functionally related L-lactate dehydrogenase (L-LDH) from dogfish as a template. This protein displays only 23% sequence identity to L-HicDH. The active site of L-HicDH was modelled by homology to the L-LDH based on the conservation of catalytically essential residues. Substitutions of the active site residues Gly234, Gly235, Phe236, Leu239 and Thr245 were made in order to identify their unique participation in substrate recognition and orientation. The kinetic properties of the L239A, L239M, L236V and T245A enzyme variants confirmed the structural model of the active site of L-HicDH. The substrates 2-oxocaproate, 2-oxoisocaproate, phenylpyruvate, phenylglyoxylate, keto-tert-leucine and pyruvate were fitted into the active site of the subsequently refined model. In order to design dehydrogenases with an improved substrate specificity towards keto acids branched at C3 or C4, amino acid substitutions at positions Leu239, Phe236 and Thr245 were introduced and resulted in mutant enzymes with completely different substrate specificities. The substitution T245A resulted in a relative shift of substrate specificity for keto-tert-leucine of more than 17000 compared with the 2-oxocaproate (kcat/KM). For the substrates branched at C4 a relative shift of up to 500 was obtained for several enzyme variants. A total of nine mutations were introduced and the kinetic data for the set of six substrates were determined for each of the resulting mutant enzymes. These were compared with those of the wild-type enzyme and rationalized by the active site model of L-HicDH. An analysis of the enzyme variants provided new insight into the residues involved in substrate binding and residues of importance for the differences between LDHs and HicDH. After the protein design project was complete the X-ray structure of the enzyme was solved in our group. A comparison between the model and the experimental 3D structure proved the quality of the model. All the variants were designed, expressed and tested before the 3D structure became available.

Crystallographic and enzymatic investigations on the role of Ser558, His610, and Asn614 in the catalytic mechanism of Azotobacter vinelandii dihydrolipoamide acetyltransferase (E2p).

Dihydrolipoamide acetyltransferase (E2p) is the structural and catalytic core of the pyruvate dehydrogenase multienzyme complex. In Azotobacter vinelandii E2p, residues Ser558, His610', and Asn614' are potentially involved in transition state stabilization, proton transfer, and activation of proton transfer, respectively. Three active site mutants, S558A, H610C, and N614D, of the catalytic domain of A. vinelandii E2p were prepared by site-directed mutagenesis and enzymatically characterized. The crystal structures of the three mutants have been determined at 2.7, 2.5, and 2.6 A resolution, respectively. The S558A and H610C mutants exhibit a strongly (200-fold and 500-fold, respectively) reduced enzymatic activity whereas the substitution of Asn614' by aspartate results in a moderate (9-fold) reduced activity. The decrease in enzymatic activity of the S558A and H610C mutants is solely due to the absence of the hydroxyl and imidazole side chains, respectively, and not due to major conformational rearrangements of the protein. Furthermore the sulfhydryl group of Cys610' is reoriented, resulting in a completely buried side chain which is quite different from the solvent-exposed imidazole group of His610' in the wild-type enzyme. The presence of Asn614' in A. vinelandii E2p is exceptional since all other 18 known dihydrolipoamide acyltransferase sequences contain an aspartate in this position. We observe no difference in conformation of Asp614' in the N614D mutant structure compared with the conformation of Asn614' in the wild-type enzyme. Detailed analysis of all available structures and sequences suggests two classes of acetyltransferases: one class with a catalytically essential His-Asn pair and one with a His-Asp-Arg triad as present in chloramphenicol acetyltransferase [Leslie, A. G. W. (1990) J. Mol. Biol. 213, 167-186] and in the proposed active site models of Escherichia coli and yeast E2p.

Structure/activity relationship of adenine-modified NAD derivatives with respect to porcine heart lactate dehydrogenase isozyme H4 simulated with molecular mechanics.

Using a significantly simplified modification procedure, four charged analogues of the coenzyme NAD, N(1)- and N6-(2-hydroxy-3-trimethylammoniumpropyl)-NAD, N(1)- and N6-(3-sulfopropyl)-NAD were prepared. The kinetic parameters of these derivatives and N(1)-(2-aminoethyl)-NAD, N6-(2-aminoethyl)-NAD and tricyclic 1,N6-ethanoadenine-NAD, all with alterations to the adenine moiety, were determined for porcine heart lactate dehydrogenase isoenzyme H4. The coenzyme activity depends on both position and charge of the introduced groups. Modification of the N6-position leads to a 25-250-fold increase of the kcat/Km value compared to the related N(1) derivative. The kcat/Km value for 1,N6-ethanoadenine-NAD is in the range between that of N(1)-(2-aminoethyl)-NAD and N6-(2-aminoethyl)-NAD. In the case of both N(1) and N6 functionalization, the Km values increase from (3-sulfopropyl)-NAD, with a negatively charged substituent at the adenine, over (2-amino-ethyl)-NAD to (2-hydroxy-3-trimethylammoniumpropyl)-NAD with an uncharged and positively charged substituent, respectively, at the adenine. All N6 derivatives are analogues like NAD with respect to Km and/or Vmax and kcat/Km. The conformation of NAD and its derivatives was calculated and their interaction in the active site of lactate dehydrogenase was simulated using the molecular mechanics program AMBER. The significant differences in activity in correlation to porcine heart lactate dehydrogenase isoenzyme H4 could be rationalized by modelling the three-dimensional structure of the NAD site.

Crystal structure of glucose oxidase from Aspergillus niger refined at 2.3 A resolution.

Glucose oxidase (beta-D-glucose: oxygen 1-oxidoreductase, EC 1.1.3.4) is an FAD-dependent enzyme that catalyzes the oxidation of beta-D-glucose by molecular oxygen. The crystal structure of the partially deglycosylated enzyme from Aspergillus niger has been determined by isomorphous replacement and refined to 2.3 A resolution. The final crystallographic R-value is 18.1% for reflections between 10.0 and 2.3 A resolution. The refined model includes 580 amino acid residues, the FAD cofactor, six N-acetylglucosamine residues, three mannose residues and 152 solvent molecules. The FAD-binding domain is topologically very similar to other FAD-binding proteins. The substrate-binding domain is formed from non-continuous segments of sequence and is characterized by a deep pocket. One side of this pocket is formed by a six-stranded antiparallel beta-sheet with the flavin ring system of FAD located at the bottom of the pocket on the opposite side. Part of the entrance to the active site pocket is at the interface to the second subunit of the dimeric enzyme and is formed by a 20-residue lid, which in addition covers parts of the FAD-binding site. The carbohydrate moiety attached to Asn89 at the tip of this lid forms a link between the subunits of the dimer.

Crystallization and preliminary X-ray diffraction studies of a deglycosylated glucose oxidase from Penicillium amagasakiense.

The dimeric glucose oxidase from Penicillium amagasakiense was deglycosylated, purified and crystallized as a complex with its coenzyme FAD. Deglycosylation and purification to isoelectric homogeneity were shown to be an important prerequisite step to obtain crystals suitable for X-ray investigations. Crystals of the deglycosylated enzyme were reproducibly grown using ammonium sulfate as precipitant at pH 7.4 to 7.5. Crystals diffract to at least 2.0 A resolution and belong to the orthorhombic space group P2(1)2(1)2(1), with refined lattice constants of a = 59.3 A, b = 136.3 A and c = 156.7 A. Assuming two monomers (approximately 135 kDa) per asymmetric unit the Vm value is 2.3 A3/Da.

Purification of the glycoprotein glucose oxidase from Penicillium amagasakiense by high-performance liquid chromatography.

Fast protein liquid chromatography (FPLC) in combination with ion-exchange chromatography on a Mono Q column was used to purify glucose oxidase from Penicillium amagasakiense to homogeneity. Purification was performed with a mixed pH and salt gradient, with 20 mM phosphate buffer (pH 8.5) as starting buffer (A) and 50 mM acetate buffer (pH 3.6) with 0.1 M NaCl as elution buffer (B). Elution conditions were optimized to permit the simultaneous purification and separation of the glucose oxidase isoforms. Three peaks, each consisting of 1-2 isoforms and exhibiting a homogeneous titration curve profile, were resolved with a very flat linear gradient of 5.0-5.1% B in 40 ml. Three more peaks, each consisting of several isoforms, were eluted at 10%, 30% and 100% B. Optimization of the elution conditions and separation of the glucose oxidase isoforms was only possible because of the rapidity of each purification step and the high resolution provided by FPLC and Mono Q.